'1
"If.“ f

M ‘ . n.
'6‘ ‘;":‘.¢1 712“?“
, ‘332‘55 5’3“
’E 3? ‘1'

o 8,. I
‘. uI'FLP—gsr

_ *;"Ha=‘~tr-.-. '
...-2"”: 2.; L‘ ' "
'7'»-

n; ‘ ‘u I
'JII"I+~
. I II. ._I
I 'EJIEII'I 0v‘53{;1_\
“ ‘l

.‘n“,
‘7 In".

. |;.IT‘..
Hm. I'J’E~“"IIII
.::;.7|“ .' :If!
' IIIT"

:II‘II II bu"; .I ‘, .. .o‘.. ,1. . . I" I?" - I MI 2".“ ‘ , I ""P‘.
4-..} 1'54“ -.  ‘ ‘31,.“ I. If” j. ... .V’ .. ..n 3")” .' -}- '- ‘

‘EIIrr PI “Itiflxh‘

I f,‘ 'I‘ O .n‘II:IYtL?

I "In“ A? (III:
'ytl’ ”if“ ' lul-

 

!

‘ I‘I “I I

II" I ’fl'fk
II‘: ”I" ":3

 

..
I b
I.”

This is to certify that the

thesis entitled

SOME KINETIC ASPECTS OF THE GASIFICATION
OF GRAPHITE AND PLATINUM GRAPHIMET

presented by
Seung I. Choi

has been accepted towards fulfillment
of the requirements for

M. S. dcgnfi3h1 Chemical Engineering

W }//D¢ Wm

/ Major professor

 

0-7639 MS U is an Affirmative Action/Equal Opportunity Institution

 

 

)VIESI.) RETURNING MATERIALS:
Place in book drop to
LIBRARIES remove this checkout from
g—c— your record. FINES will

 

 

 

be charged if book is
returned after the date
stamped below.

 

 

 

 

SOME KINETIC ASPECTS OF THE GASIFICATION
OF GRAPHITE AND PLATINUM GRAPHIMET

BY

Seung I. Choi

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF SCIENCE

Department of Chemical Engineering

1984

Tl \\

ABSTRACT

SOME KINETIC ASPECTS OF THE GASIFICATION
OF GRAPHITE AND PLATINUM GRAPHIMET

BY

Seung I. Choi

The kinetics of the oxidation and steam gasification of

graphite and platinum graphimet is presented. The kinetic
studies employ temperature programmed desorption/reaction.
A scanning transmission electron microscopy of both unreacted
and reacted platinum graphimet reveals that the behavior and
morphologies are similar to those found by controlled atmos-
phere electron microscopy.

A correlation using a mathematical model of the temper-
ature programmed desorption/reaction experiments show that
platinum increases the density of active sites. This is at‘
tributed to the scission of carbon-carbon bonds by platinum.
The lowering of the catalyzed oxidation temperature is ex-
plained by the scission function of platinum. For platinum,
there is some possibility that both electron and oxygen
transfer could be operating for the catalyzed reaction. The
latter is suggested by a match between Baker's controlled
atmosphere electron microscopy studies and a simple mathema—
tical model using the assumptions of viscous oxide layer and

surface diffusion.

TABLE OF CONTENTS

Page
List of Figures iv
Notations vii
CHAPTER I
INTRODUCTION 1
LITERATURE REVIEW
Mechanism of Oxidation 3
Temperature Programmed Desorption 6
OBJECTIVE OF RESEARCH 9
CHAPTER II
THE TEMPERATURE PROGRAMMED DESORPTION/
REACTION APPROACH OF THE GASIFICATION
OF CARBON: EXPERIMENT
Experimental Method 11
Experimental Results 17
Discussions on Results 37
CHAPTER III
THE TEMPERATURE PROGRAMMED DESORPTION/
REACTION APPROACH OF THE GASIFICATION
OF CARBON: THEORY
Theory 44
Theoretical TPD Plots 51

Discussion on TPD Plots 59

ii

CHAPTER IV
THE CHANNEL PROPAGATION RATE OF PLATINUM
ON A CARBON SURFACE: A MODEL AND STEM
STUDIES
Background

STEM of Unreacted and
Reacted Platinum Graphimet

A Mathematical Model of the
Catalyst Particle Propagation

Closures

APPENDIX
A. Computer Program and Sample Calculations
B. Standarization of G. C. Units

C. Solutions to Equations (IV - I) to (IV - 4)

LIST OF REFERENCES

64

67

89
100

101
105

112

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

II—4A

II-4B

II-5A

II-5B

II-6A

II-GB

II-7A

II-7B

II-SA

II-8B

II-9A

LIST OF FIGURES

page
Schematic diagram of the experimental
apparatus ............. . .................. 12
Schematic sketch inside the reactor ...... l3
Desorption pattern of graphite pre-
treatment, fi= 30. 5° C/3 min .............. 17
Weight loss pattern of graphite pre-
treatment, fl= 30.5 C/3 min .............. 18
Desorption pattern of graphite
second heating, fi= 30. 5 C/3 min ........ . 19
Weight loss pattern of graphite
second heating, fi= 30. 5C/3 min ......... 20
Desorption pattern of graphite
third heating, fi=—29.5 C/3 min .......... 21
Weight loss pattern of graphite
third heating, p: 29.5 C/3 min .......... 22
Desorption pattern of graphite
fourth heating, ,3: 31 C/3 min ........... 23
Weight loss pattern of graphite
fourth heating, fl= 31° C/3 min ..... 24
Desorption pattern of Pt- -graphimet
pretreatment, 5= 30’ C/3 min ............. 25
Weight loss pattern of Pt- -graphimet
pretreatment, p: 30 C/3 min ............. 26
Desorption pattern of Pt-graphimet
second heating, fl: 27 C/3 min ........... 27
Weight loss pattern of Pt— graphimet
second heating, 3: 27° C/3 min ..... 28

Desorption pattern of Pt- graphimet
third heating, p= 31° C/3 min ..... 29

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

II-9B

II-lOA

II-lOB

II-ll

II-12

II-13A

II-13B

III-l

III-2

III-3

III-4

III-5

III-6

III-7

Weight loss pattern of Pt- graphimet
third heating, fi = 31 C/3 min ...........

Desorption pattern of graphie
H20 adsorption, '8 = 29. 5 C/3 min ........

Weight loss pattern of graphite
H20 adsorption, fl = 29. 5° C/3 min ........

Desorption pattern of 8Pt—graphimet
HZO adsorption, fl = .3 C/3 min ........

Steam gasification of graphimet
TPD, p: 23° C/3 min .....................

Steam gasification of Pt— graphimet
TPD, p = 29.4 C/3 min ...................

Steam gasificationof Pt- graphimet
weight loss pattern, 19‘ 29. 4 °C/3 min

Schematic sketch of TPD system ..........

Theoretical TPD spectra of carbon/
metal system, EC = 260 KJ/mole,
Em = 230 KJ/mole .... ....................

Theoretical TPD spectra of carbon/
metal system, EC = 260 KJ/mole,
Em = 250 KJ/mOle ... ....... . .............

Theoretical TPD spectra of carbon/
metal system, Ec = 260 KJ/mole,
Em = 270 KJ/mole ..... ...................

Theoretical TPD spectra of carbon/
metal system, EC = 260 KJ/mole,
Em = 280 KJ/mole .. ........ . ............ .

Theoretical TPD spectra of carbon/
metal system, EC = 260 KJ/mole,
Em = 300 KJ/mole ........................

Theoretical TPD spectra of carbon/
metal system, Ec = 270 KJ/mole,
Em = 250 KJ/mole ........................

page

51

52

53

54

55

56

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

Figure

III-8

III-9

IV-1

IV-2

IV-3

IV—4

IV—5

IVb6

IV-7

IV-8

IV-9

IV-lO

IV-ll

IV-12

IV-l3

page

Theoretical TPD spectra of carbon/
metal system, EC = 270 KJ/mole,
Em = 300 KJ/mole ................ ........ 57

Theoretical TPD spectra of carbon/

metal system, EC = 270 KJ/mole,

Em = 350 KJ/mole ........................ 58

STEM of unreacted Pt- graphimet
1, 000 A/cm .............................. 71

STEM of unreacted Pt- graphimet
500 A/cm ............... . ............... 72

STEM of unreacted Pt- graphimet
50 A/cm ................................. 73
20 A/cm ................................ 74

X-ray dispersive analysis of
Pt-graphimet ... ......................... 75

STEM of reacted Pt-graphimet
2, 000 A/cm ................ ..... ......... 76

STEM of reacted Pt— graphimet
1,000 A/cm .............................. 77

STEM.of reacted Pt-graphimet
500 A/cm ............... . ................ 78
100 A/cm ....... . . .... ................. 79

Schematic illustration of various
channelling activities in Figure (IV-7).. 80

Simplified sketch of particle
channelling activity - model .. ........ .. 88

Dependency of channel propagation
rate on particle width .................. 98

Dependengy of channel propatation
rate on k ............ . ................. 99

vi

U

C -mo

CS

W??? X‘ D" '11

NOTATIONS

Preexponential factor in the carbon phase

( sec-l
Preexponential factor in the metal phase -1

( sec
Concentration of carbon atoms ( moles/cm3
Concentration of carbon monoxide ( moles/cm3
Concentration of carbon atoms at the carbon- 3
metal interface ( moles/cm
C / Co
1-2

Diffusivity of carbon atoms in the metal matrix
( cm2/sec

Diffusivity of carbon atoms in the metal oxide
layer ( cmZ/sec

v

Vvv

v

v

)

Diffusivity of carbon atoms on the surface of the

metal oxide layer ( cmZ/sec

Activation energy of desorption in the carbon

phase ( KJ/mole
Activation energy of desorption of the metal

phase ( KJ/mole
Flow rate of carbon monoxide ( moles/sec
Plank constant ( 6.626 x 10-34 J sec
Boltzmann's constant ( 1.380 x 10—23 J/K
First order rate constant ( sec-1

A thermodynamic constant defined in page 92 —l
( cm

)

V

vvvvv

s-bww

5U N O :1 5’2

0']

Mass transfer coefficient of carbon atoms

( cm/sec )
AC exp(47é) ( sec.l )
Am exp(4xn) ( sec—l )
The metal particle width ( cm )
Molar flow rate ( moles/sec )
Flux of carbon atoms ( moles/cm2 sec )

Number of moles of carbon monoxide per unit area
( moles/cm2 )

Number of moles of carbon monoxide per unit

volume ( moles/cm )
Integers

Flow rate of carrier gas ( moles/sec )
Gas constant ( 3.314 J/gmoleoK )
Rate of carbon monoxide production ( moles/cm2 )
Total effective surface area ( cm2 )
Effective surface area of carbon phase ( cm2 )
Effective surface area of metal phase ( cm2 )
Sm / SC

Temperature of TPD system 1 (OK )
Reference temperature ( 298'K )
Maximum peak temperature of TPD spectra (°K )

Temperature of the inflection point of TPD spectra
( K )

c c
nco / nco (l)
m m

nco / nco (1)

viii

v T / TO

(V) Velocity of channel propagation
x,y,z Cartesian coordinates
Superscripts
m Metal phase
c Carbon phase
5 Surface of metal oxide layer
Subscripts
o Metal-carbon interface

Greek Symbols

Heating rate
Ec / RTo

1b

Em / RTo

p/%

x / L

y/L

Biot number = Kc L / Dcm
k1 L2 / Dcs

S

D L Co /,Dcs Co

cm
Effective thickness of oxide layer

s s
C (0) / CD

~ 5
Dc-mo k Co L / Dcm Co 6

£S~ >Oo;e_;9~;e~u~<mm 573;?

ix

( cm/sec )

( K/sec )

1 )

( sec-

(cm)

 

,
‘-
L.
.L'
.-‘
v. -—
9“. ':-t:‘ I
o
‘0 .
‘ .', l .

, .

It 1.; 1r

 

CHAPTER I

 

INTRODUCTION

Catalytic oxidation of carbon and graphite is one of the
most interesting and intriguing topics in the field of cataly-
sis. It is well known that extremely small amounts of impuri—
ties can have a profound effect on the oxidation rate of graph—
ite, but in spite of considerable efforts made to this topic
over the years the precise mechanism of how impurities operate
is still unknown.

Of many important steps involved in the gasification of
graphite, the breakage of the carbon—carbon bonds by catalyst,
the adsorption of reacting gases on the surface of catalyst,
and the desorption of products from the surface seem to be the
most significant ones, and most works in the literature have
been directed toward studies on the characterization of the
catalytic surfaces and the determination of the kinetic para-
meters of both the overall and each of the three steps men-
tioned above.

The interaction of platinum group metals with graphite is
of particular importance since these metals are the catalysts
for a number of reactions of commercial interest, e.g., reform-
ing of naphthas and certain hydrogenations, and an oxidation
steps is usually employed to reactivate a carbon contaminated
catalyst.

It is the purpose of this research to examine the details

     

     

gather: with the use of Temperature Programmed Desorption and
" "’7fleanning Transmission Electron Microscopy and to pursue an
‘5 fiunderatanding of the catalytic mechanism, especially of the

1"fixidation of carbon.

LITERATURE REVIEW

A. Mechanism of Oxidation of Carbon

While there has been a surge in the number of experiments
on metal-catalyzed oxidation of carbon, only a few have attem—
pted to offer the mechanism of the oxidation process. This ad-
versity is mainly attributed to the paucity of information
regarding the mechanism. There are two mechanisms that have
been used to explain the activity of metal catalysts in carbon
oxidation. In the electron transfer mechanism (11, 12), the
catalyst serves to either donate or accept electrons to or from
the carbon. With the weakening of the carbon—carbon bond on
the edge planes of carbon, the dissociative chemisorption of
oxygen is greatly enhanced. In oxygen transfer mechanism (13),
~ the active catalyst is considered as an oxygen carrier which
perhaps undergoes a periodic number of oxidation-reduction
cycles on a carbon surface during the progress of reactions.
An intermediate compound such as a metal oxide is formed, and
is redUCed upon contact with the carbon substrate at the metal"
carbon interface. Recently, the catalyed carbon-carbon bond
breakage mechanism has been proposed for carbon gasification
by H

H O and CO2 by Holstein and Boudart (14, 15). They

2’ 2
stress that the ability of the catalyst to directly break
carbon—carbon bonds without the participation of adsorbed gas
on the catalyst surface, and the catalyst must also be capable

of dissociatively adsorbing gas molecules which react with the

adsorbed carbon on the catalyst surface.

It now appears as a consensus that a whole set of physico-
chemical factors govern the rates of metal/metal oxide catalyzed
oxidation of carbon on a graphite basal surface.

Two of these factors are effective catalyst particle size
and the oxidation state of the metal. The particle size of the
metals and metal oxides often dictates the magnitude of the rate
of carbon oxidation, and sometimes it is generally adopted as a
characterization of metal/metal oxide catalytic oxidation of
carbon. As in most of Baker's and his co-workers studies, the
channel propagation rate of a single catalyst particle is always
adopted as the basis for estimating the magnitude of the cataly-
tic attack rate of graphite surfaces. The channel propagation
rate, as observed by Controlled Atmosphere Electron Microscope
(CAEM), is inversely proportional to the particle size, ex-
pressed in terms of particle width, for the oxidation process
(16, 17, 18, 19, 20). On the contrary, for the metal catalyzed
hydrogenation of graphite surfaces, the channel propatation rate
has a quadratic dependency (17). Baker and Sherwood (17) there-
fore conclude that there exists a fundamental difference in the
mechanisms of carbon oxidation and hydrogenation. Keep et a1.
(21) have proposed two mechanisms for catalytic reduction of
graphite. In the first mechanism, the metal catalyst acts as
a dissociation center for hydrogen, producing atoms which dif-
fuse across the metal/carbon interface and react with the car-
bon to produce methane. In the second mechanism, carbon atoms

at the interface dissolve into the metal and diffuse to the

exposed particle surfaces to react with hydrogen. Baker and
Sherwood (17) support the first mechanism as a more rational
model from their investigations involving silica inclusions in
small nickel particles, and they further conclude that the met—
al catalyzed oxidationci graphite is also diffusion-controlled.

Holstein and Boudart (14), in their study of the platinum
catalyzed hydrogenation of graphite, argued that hydrogen spill—
over, and thus, surface diffusion was not a viable mechanism
and could not be a limiting process. They hypothesized that
the cleavage of carbon-carbon bonds by the metal could be the
limiting process.

Another important aspect of the metal-catalyzed oxidation
of carbon is the chemical state of the metal. The oxidation
state of the catalyst has been found to have a significant ef-
fect on the rate of carbon oxidation. For instance, iron is
very active in the CO2 oxidation of carbon, Wustite (FeO) is
less active and magnetite (Fe304) is totally inactive (13).

In the oxidation of graphite by Ni, Baker and Sherwood (l7)
speculated that a thin oxide layer of NiO was responsible for
the passivism of the nickel catalyst towards C-O2 reaction.

For such metals as Ag, Pb, and Pt, the rate of the catalyzed
oxidation 01 carbon is not drastically influenced by their oxi-
dation states. The observation of the dynamics of metal oxides
formation during the course of carbon oxidation is yet beyond
in situ analysis, and the speculation of metal oxide formation

is usually done by inference.

B. Temperature Programmed Desorption

Various techniques are used to study the gas-solid surface
interactions, such as flash—filament desorption, selective poi-
soning, field electron and ion microscopy, etc. Temperature
Programmed Desorption (TPD) method is in principle similar to
flash-filament desorption, but TPD method can be applied to any
type of solid catalysts and situations where the experimental
conditions need be closer to those in actual practice. A de—
tailed description of TPD and its applications is given by
Cvetanovic and Amenomiya (1, 2).

The simplest model by Cvetanovic and Amenomiya (1) assumes
linear heating schedule, first-order desorption, homogeneous
surface, negligible diffusional effects and neligible readsorp-
tion, and gives the following equation relating the maximum peak

temperature to the apparent activation energy of desorption.
2 1n(TM) — 1n(p) = Ed/RTM + 1n(Ed/AR) ...... .....(1-1)

When readsorption occurs freely during desorption, the
peaks are much broader, and readsorption can be minimized by
increasing the carrier-gas flow rate. Unlike the theoretical
arguments by Cvetanovic and Amenomiya, Gorte (3) argues that
readsorption in porous catalysts cannot be eliminated even for
desorption into a vacuum, and readsorption can easily change
the desorption temperature by several hundred K. Hertz et a1
(4) studied desorption of carbon monoxide into a vacuum for a
1.0 wt % platinum on porous 'r- A1203,
sults with those by Foger and Anderson (5) who did the same

and compared their re-

experiment using a carrier gas instead of vacuum. They found
that the peak temperature in the case of vacuum desorption was
about 869K higher than that in the case of desorption into a
carrier gas, and in both cases adsorption competed with desorp-
tion to the extent that adsorption equilibrium was closely ap—
proached at each point in a porous sample.

In the case of desorption from heterogeneous surfaces,
theoretical analyses of TPD model become more elaborate.
Cvetanovic and Amenomiya (2) used the Arrhenius equation to de—
fine the dependence of rate constant on temperature, and assumed
that the activation energy of desorption was a linear function
of surface coverage. They found from calculations that TPD
spectra had a broad energy spread on a heterogeneous surface,
and as a potential example showed the desorption spectrum of
carbon monoxide from a sample of plantinum deposited on cabosil.
Taylor and Weinburg (6) developed a model for analyzing TPD
spectra to determine the coverage dependence of the preexponen-
tal factor and the desorption energy in the Arrhenius equation
for a one—step desorption process. They showed that both rate
parameters had a strong coverage dependence in the case of car-
bon monoxide desorption from Ir (110), and concluded that among
several limitations their model was not appropriate for systems
involving desorption from several distinct sites. Barton and
Harrison (7) studied desorption of surface oxides after chemi-
sorption of oxygen on graphite, and found that desorption of
the oxides formed during chemisorption was mainly in the form

of carbon monoxide and the surface was reasonably uniform with

an activation energy of desorption of 276 KJ/mole. They further
concluded that the activation energies of oxidation and those of
desorption of oxide were so close that the oxidation of graphite
proceeded in two steps, formation of surface oxide and desorption
of this oxide, and the desorption could be rate controlling.

They also found that the amount of oxide desorbed after chemi-
sorption was substantially less than that found during the ini-
tial degassing, corresponding to oxide areas of 9.0 m2/g and

6.4 mZ/g respectively. Although they did not carry out the third
heating of the same sample after the second chemisorption of
oxygen to make this trend clearer, they still concluded that the
heat treatment during the initial degassing rendered part of the
graphite surface incapable of reforming oxide. Other workers
have explained this. Harker et a1. (8) interpreted the reduced
reactivity in terms of the annealing of basal plane dislocations,
while Mrozowski and Andrews (9) interpreted their results by
assuming that free valencies caused by the grinding process were
capable of bonding with adjacent free valencies leaving a chemi-
cally inactive surface.

When desorbing species has an appreciable surface reaction
rate during desorption, Gorte and Schmidt (10) shows that reac—
tion parameters can only be obtained from TPD experiments when
reaction occurs in the same temperature range as desorption,
and in the limit of low coverage one rate parameter kinetics
may be applicable in most systems with heating rate being the
most useful variable in extracting reaction parameters. They

also conclude that the desorption peaks shift to higher temper-

atures and are narrower than would be predicted by their desorp-

tion parameters alone.

OBJECTIVE OF THE RESEARCH

This work reexamine the platinum catalyzed oxidation of
carbon, with emphasis on understanding the underlying kinetics
and catalytic function of platinum. We aim to investigate
the following:

(1) platinum as an agent for carbon-carbon bond scission

(2) oxygen and electron transfer mechanism

(3) existence of viscous layer of metal oxide

(4) the feasibility of employing temperature programmed

desorption/reaction in studying the kinetics of catal-
yzed and uncatalyzed oxidation of carbon

(5) assess the catalytic behavior of platinum graphimet

using scanning transmission electron microscopy.

CHAPTER II

THE TEMPERATURE PROGRAMMED DESORPTION/REACTION
APPROACH OF THE GASIFICATION OF CARBON:

EXPERIMENT

ll

EXPE RI MENTAL METHOD

This chapter discusses the temperature programmed desorp-
tion/reaction experiments for the oxidation of graphite and

platinum graphimet.

A. Design of the Apparatus

A schematic diagram of the experimental set-up is given
in Figure (II-1).

The quartz reactor tube was 30 inches long and 1.5 inch
I.D. The sample of catalyst powder (typically 50 mg) was sus-
pended from a Cahn 1000 microbalance with a quartz fiber (4
feet long and 500 microns in diameter) into the center of the
quartz tube reactor which was surrounded by 1400 watts Lindberg
tubular turnace of 2 inches I.D. The temperature of the fur—
nace was varied by Lindberg Furnace Controller of the propor-
tional integral type. Chromel-Alumel thermocouple enclosed in a
thin quartz tube was placed right next to the quartz sample pan
for the temperature measurements of the sample. The voltage
readings were monitored by a potentiometer, and converted into
degrees Fahrenheit using a conversion table provided by Omega

Corporation. (For details of the reactor, see Figure (II-2).

B. Catalyst
The catalyst used throughout the experiment was a 1 wt. %
platinum graphimet which is commercially produced by the Ventron

Corporation. Although graphite cannot be directly intercalated

12

Amman ocflumos "m .wocmHmnouoflE “2
.Hmaooo “U .mHQSOUOEuwzu “B .umquoHucwaom "m .momcusw "m
.Houomwu “m .uoumusuom 0N3 "m .mummemuou "Em .m>am> wapwmc ">v

wouMHMQd4 Hmpcmefluomxm mo Emndmao ofiumfiosom HIHH wusmflm

 

‘ I

N
o: > O

 

 

(IV h2w>

 

 

 

 

l3

 

 

 

 

 

 

Figure II-2 Schematic Sketch Inside the Reactor
(a: reactor wall, b: quartz fiber,

c: quartz pan, d: sample,
e: quartz tube for thermocouple)

with transition metal, intercalation is claimed to be possible
by the reduction of intercalated salts of these metals. A de-
tailed description of the preparation is given in a U.S. patent
disclosure by Lalancette (29). On the basis of the total metal
content in graphimets, the distance between each metal atom is
calculated to be on the order of 5A (30). A highly purified
graphite powder supplied by Ultra Carbon Corporation (Type UCP-

2, 325 mesh) was used as a reference for the catalytic study.

14

C. Experimental Procedure
1. Pretreatment of Catalysts

Each fresh sample out of the bottle was outgassed at one
atmospheric pressure from room temperature to 1,000.C. Linear
temperature programming was obtained manually by closely moni-
toring the temperatures of the sample within the reactor. The
carrier gas used throughout the experiment was helium of high
purity (>-99.99%), and the flow rate was set at 200 ml/min.
0.4 ml of desorbing gas was collected into a high precision
syringe from the sample collecting flask every three minute
and injected into the Varian 3700 Gas Chromatograph equipped
with a thermal conductivity detector for gas analysis. Con—
trolled volumes of reference gases of known concentrations
were also injected into the gas Chromatograph for standardiza-
tion and quantification of sample gases from experiments. The
standardization procedure is given in Appendix.

The signals of gas Chromatograph of products were con-
verted into concentrations according to the standardization,
and the concentrations in gram moles were plotted as a function

of temperatures in Celsius.

2. Controlled Adsorption
After pretreatment in helium up to 1,000°C, the sample was
cooled down to room temperature at atmospheric pressure with
helium gas continuously flowing through the system during the
entire cooling period. During controlled oxygen chemisorption,

surface oxides were formed at 25'C by exposing the pretreated

15

sample to an oxygen pressure of 20 psig for 30 minutes.

H 0 adsorption was performed at 300°C by flowing helium

2

. 0 .
gas which was saturated w1th H20 at 25 C 1n the H20 saturator

into the system for 30 minutes.

3. Temperature Programmed Desorption

After the controlled adsorption, the sample was outgassed
in helium at a flow rate greater than 500 ml/min for 20 minutes
at the adsorption temperature to remove any traces of oxygen
in the system. The methods of linear temperature programming
and sample collection were the same as in the pretreatment.
Concentrations of desorption products, C0, C02, CH4, and H2 were
expressed in terms of gram moles and plotted as a function of
temperature. Since the sample size was 0.4 ml, and the carrier
gas flow rate was 200 ml/min, the concentrations in gram moles
can be converted to the rate of total production in gram moles/
min by multiplying them by a factor of 500. After the outgass-
ing up to 1000°C, the sample was again cooled down to room tem-
perature in the presence of flowing helium, and the same pro-
cedures were repeated for controlled adsorption and linear tem—
perature programmed heating to obtain second and third concen-
tration versus temperature plots. After the third heating, the
sample was cooled to room temperature, and collected for scan-

ning transmission electron microscopic analysis.

4. Gasification and Oxidation

Steam gasification was performed with a fresh sample.

l6

Helium gas saturated with H20 at room temperature was contin-
uously fed into the system at 200 ml/min, and the temperature
was raised from room temperature up to 1,000°C in an approxi—
mately linear fashion.

Oxidation was performed isothermally at temperatures above
500°C with helium at a rate of 200 ml/min, and the gas sample
were collected for analysis after a steady weight decrease was

observed on the recorder for at least 10 minutes.

 

EXPERIMENTAL RESULTS

17

U ooo.H
0

com

CT: m\Uom.om n1 .ucwEummuuwnm
in: Now .10c 00

wuflcmouu mo cuwuumm coflpduOmoo <mIHH musmfim

 

 

 

com com oow
—

 

 

mloax
mmHOEm

cflE m\Uom.om nu .ucwfiummuumum

wuflsmmuu mo cumuumm mwoq unmfimz MMIHH musqfim

moo.H mmm mmm vmm mmh wow owm mom va mmm

 

18

 

. . _ _ . ,_ _ a _

 

 

 

19

o ooo.H
0

Uomm um musmoaxm No CHE om
:flE m\Uom.om M Q .mcwuww: pcoomm
luv Noo .loc oo

oufizmmuo mo cuwuumm :ofiuQuOmmo

com com

«vIHH musmflm

con

cow

 

 

 

 

mica x
wHoEm

20

Uomm um wusmomxw No use om
CHE m\Uom.om u h .mcflummm pcoowm

ouflsmmuo wo :nwuumm mmoq usmwmz

Uo wmm Nam mwm hmh wmh

 

menHH wnsmam

vmw cam 00

 

      

.
u
.

1 mg

 

,-.,._._ ——

 

 

 

 

21

00mm um wusmomxm No :fiE om

fie Qoomam u m. .3335 Eda
in: N8 Loo oo

wuflnmmuw mo cumuumd :ofiuQuOmwQ mmIHH wusmfim

 

 

 

 

U coo.H com com con com W
o ((+( _ J _ a O
U n .

mloa x

wHOEm

22

0 mm um wnsmomxm No ewe om

O

fie mxoomam um. 6:38: 32.2.

unflzmmno mo cumuumm mmoq unmask

mmIHH musmflm

 

 

 

 

 

23

00000 H

09mm um wuomomxw NO aflE om
cue mxooam n.m .mcaummm sunset

in; N8 .318

wuflsmmuo mo cuwuumm coaumuowwo

com com

<muHH museum

 

 

con
_

oow
_

 

 

 

mica x
OHOEm

24

Oomm um musmomxw No GAE om
CHE m\Uon n h‘.m:Hummm nuuoom

wuflnmmuw mo cumuumm mmoq named:

00 mmm Hmm mom Hmm vmh mow
_ _ _ _ _

 

mwaHH musmam

mam mwv
_

 

 

-

mg

 

 

 

 

25

ooo.a

swE m\ooom u fi .ucwEumwuumum

ADV

N

00

.lo 1 oo

quHnQouOIum wo :uwuumm :ofluQMOmmo

com

com

com

«FIHH musmflm

oom

 

 

 

com
d

 

 

mica x
wHOEm

26

GAE m\Uoom u ms. .ucmEumwhummm

qufinmmHOIum mo cumuumm moon “some: manH wusmflm

Uo MHo~H me omm mom New no mom HNm

 

 

 

_ . _ . _ .fi _. m a

6m

 

 

 

27

U.

0 mm on whomomxm N
O

0 use em .:22 mxooo.am ".u
.02flumm: ocoowm .ADV

Noo .loo oo

quflsmmu0|um mo cuwuumm :ofluQMOwwo dmIHH mpomflm

ooo.a com com
d - d

oon com com
-

 

 

UIIIJVIU\\\|fi1lfl1JHIIU

 

 

 

28

00mm um wusmomxw NO :flE om
CHE m\Ooo.nm n_h .mcfiummm osooom

uwfiflsamuonum mo cumuumm wmoq unmflmz

00 «mm mmm mmm Hmm Nmm one How

QwIHH OHDmHm

mmo

 

_ — __ E q. q .—

..

‘l

x

(

l

.

|
.

 

mam
.

 

 

 

29

0 mm um wusmomxo mo of: om .caE m\UoO.Hm Mn
0 .mcflpmmm menace . An: NOU . A 0v 00
quflcmmsolum mo cumuuom coflummowmo <a|HH wusmflm

 

Us ooo.H oom oow ooe oow oom
a q d

 

 

 

 

¢IOH x
mace—m

30

Uomm um wusmomxm No cwE cm
is mxoooém u m .mcflmmm one?

unswsaouolum mo cumuumm mmog unmwmz mmIHH wusmwm

Do mao.H omm vmm oam mmm own man one mom

 

   

_ _ . . _ _ _ _ _

 

 

 

 

is Qoomdm “Q
0 com um wusmomxw Omm :HE om
0

inc Noo .Aoo ou

mpflsmmuw mo cuwuumm coflumHOmmD

o o com com

doHIHH madman

com

com

 

31

 

 

 

mIoH x
mHoEm

fie Qoomam us.
Uooom um musmomxo Omm cflE om

muflsmmuo mo cumuumm mmoq usmwmz

no mmm a; 5 m8

moalHH wusmflm

eve mew mom

 

32

mg

 

 

 

 

 

33

fie m\u.m.mm nu

0 com um whomOme Om: sea cm
0

A: m: in: moo .loo oo

umefisdmuwlum wo cummumm coflumuOmwo HHuHH wuooflm

Uoooo.H oom oom com com oom

 

 

 

 

34

a? m\u.mm um.
21mm .10. ~00 ..oo oo

muflsmmuo mo :oHumonHmmw Emwum NHIHH musoam

ooh

 

 

 

 

 

L cam

 

 

fie Suzi um
lot 38 .213;an N8 .2: oo
umEASQMHOIum mo cowumowmmmmo Emmum <MHIHH musmflm

com com ooe oow oom

 

35

 

 

E . i 9.0

l

o.ooa

o.omH

l

 

 

mica x
mHOEm

36

fig 2004.2 n ml

unsanmmnouum mo
coflumofiwflmmo Emmum mo cuwuumm mmoq unmflmz

omm omm mao How mob

mmanHH whoowm

Haw 5mm

th

 

 

.. _ _ _.

 

A. _

 

 

 

37

DISCUSSIONS ON EXPERIMENTAL RESULTS

A. Pretreatment

Figure (II-3A) shows the desorption patterns of CO and
CO2 from pure graphite during initial outgassing up to l,000°C.
The TPD spectrum of CO is rather broad with a maximum at 800°C,
and that of CO2 is relatively flat with no distinct peak. The
rate of CO evolution at 800°C is 1.65 x 10-6 mole/min, about

six times greater than that of CO evolution. No distinct

2
spectrum pattern and considerably small ratecfi'production seems
to justify the assumption of negligible CO2 production. As ex-
pected, the CO spectrum has only one maximum, and even at this
low product concentration a moderate readsorption, or product
retardation, in the sample can be seen to occur, since the peak
decays rather slowly after the maximum. Due to some disturb-
ances such as noises and shocks, Figure (II-3B) does not show
any significant trend of weight loss at each temperature except
an overall weight loss of 1.1 mg up to 1,000°C. Both Figure
(II-3A) and (II-BB), however, shows that a significant desorp-
tion of surface oxides begins to take place at about 600°C.

In Figure (II—7A), the desorption pattern of CO2 from
graphimet is well developed with a maximum at about 625°C,
while the CO spectrum has one maximum at about 825°C with a
rate of 3.55 x 10.6 mole/min, about eight times greater than
that of C02 from graphimet and about 2.2 times greater than

that of CO from pure graphite. The evolution of CO2 takes place

38

below 500°C, compared with 600°C for graphite.

B. Temperature Programmed Desorption

1. Graphite

Figure (II-4A) shows a very broad but symmetric peak of
CO during the second heating. The broadness of the peak and
the slow decay at higher temperature indeed suggest that the
desorption was strongly retarded by CO, as it is by H2 re-
ported by Blakely and Overholser (22). Foger and Anderson (5)
attributed the breadth of the desorption band with smaller
secondary peaks over the span of about 400°K to the presence
in the sample of a distribution of adsorption sites with dif-
ferent CO binding energies. Their explanation is reasonable
in view of the complex surface structure of supported metal
particles and the possible influence of the oxide support on
small metal particles. However, for pure graphite sample and
graphimet used in this experiment, neither such an influence
of support material on metal nor the broad spread of different
binding energies are expected to exist,and this can be easily
seen just by noting a unique and symmetric peak for pure graph-
ite.

It can also be observed from Figures (II-5A) and (II-6A)
that the peak temperatures decreased upon successive heating.
For the second heating, the actual peak temperature appeared
to occur between 775°C and 810°C, and thus, the apparent peak

temperature is estimated to be 780°C. For apparent peak temper-

39

atures for the third and fourth heating were found to be 760°C
and 750°C, respectively.

One possible explanation for this observation is that some
amount of residual oxygen atoms were left within the sample
even after the outgassing up to l,000°C probably because of the
atmospheric pressure, and the desorption of oxides would be
more diffusion controlled during the next outgassing which
could result in the lowering of the apparent activation energy
or the apparent peak temperature. In all cases, a significant
weight decrease was not observed until the temperature reached
about 640°C, 40°C higher than that in the pretreatment, and
this small increase seems to be due to the fact that the con-
trolled chemisorption for 30 minutes formed relatively stable
oxides only on the graphite surface, which can be verified by
the sharper and better defined peaks. And as shown by Hertz
et al. (4), these peak temperatures are expected to be a little
higher if the experiment were run under vacuum. Since under
vacuum the readsorption phenomenon would be less significant,
resulting in a higher peak temperature at a greater peak inten-
sity. Assuming a lowering of about 40°C of the peak tempera-
ture due to readsorption, the use of Equation (I-l) gave the
activation energy of desorption of 300 KJ/mole with a typical
The preexponential factor,

M'
A, was approximated by KT/h = 2.13 x 1013 sec-1 for the cal-

0
experimental value of 760 C for T

culation (23).

The activation energy of desorption of CO found by other

40

experimental methods ranges from 259 to 293 KJ/mole (7, 24, 25),
and the close agreement of our calculated value seems to indi-
cate that the assumptions claimed by Equation (I-l) were reason-

ably well met in our experiment.

2. Graphimet
As in the case of graphite, the total weight loss was

constant at 2.4 mg, about twice as large as that in the pre-
treatment, and at about 510°C, a significant weight decrease
began to occur, compared with 600°C in the pretreatment of
graphimet and about 640°C in the TPD of graphite shown in Fig-
ures (II-8B) and (II-9B). This lowering of temperature, about
130°C, over the graphite suggests the definite catalytic effect
of platinum in the oxidation of graphite. Another interesting
observation is that the peaks seemed to be resolved into two
secondary peaks, and the peak temperatures increased as the
heating was repeated. It is not very clear at this point
whether the resolution indicates two distinct peaks or just
one broad peak with some flat region in it, although a non-
uniform energy distribution on the surface appeared to cause
the broadness of the peak. In either case, however, the center
of the peak shifted to higher temperatures from one run to the
next, and this observation is exactly the opposite of that made
in the runs with graphite. Although it seems to be not plausi-
ble to use Equation (I-l) for the calculation of the activation
energy of desorption because of the clearly stronger hetero-

geniety of the surface of graphimet than that of graphite, the

41

activation energy of desorption of graphimet is expected to be
a little higher than that of graphite, since the apparent peak
temperature of graphimet is higher by more than 100°C. The
possibility of an increase in activation energy due to cataly-
sis has been reported to hold for carbon gasification by oxygen
(26), but whether the increase observed in our case was due to
the actual increase in the activation energy of desorption or
not is uncertain.

However, unlike in the case of graphite where surface
oxides were expected to be formed upon the adSorption of oxygen
atoms, the oxygen atoms adsorbed on the metal surface have to
combine with carbon atoms at elevated temperatures before the
desorption of products takes place. Winicur (27) found that
by using a Low-Energy Molecular-Beam-Scattering (LEMS) techni-
que, the desorption of CO from a Pt (lll) crystal was adequately
represented by a first order Arrhenius form with an activation
energy of desorption of 130. 2 KJ/mole and a preexponential
factor of 2.7 x 1013 sec—l. This low activation energy of de-
sorption clearly suggests a strong influence of the carbon—
carbon bond breakage by the metal and probably by the diffusion
limitations imposed upon the overall TPD results for graphimet.
Otto and Shelef (28) explained the increased activation ener-
gies of graphimets with burn-off % in terms of pore diffusion.
They concluded that the lower activation energies, observed
for fresh samples, were caused primarily by diffusion barriers

and as the pore structure Opened up during gasification, pore

42

diffusion became less important resulting in the increase in
activation energies. Their explanation was further confirmed
by the X-ray diffraction study, and our experimental observa-
tion of the increase in the peak temperature with repeated
beatings appears to be well explained by their argument. As

a consequence, it seems to be reasonable to relate the peak
temperature to the apparent activation energy even in the
graphimet sample as well as in the graphite and assume that

the apparent activation energy of graphimet is at least as high
as that of graphite. Therefore, the increased overall rate of
CO evolution of graphimet over that of graphite can be attri-
buted to the increase in the preexponential factor in the
Arrhenius rate expression, and this conclusion strongly suggest
that the cleavage of the carbon-carbon bonds may be the most
significant step in the metal catalyzed oxidation of carbon,

if not the rate-determining step.

The increased rate of the second and third run, as obser-
ved by the doubled weight loss compared with that in the pre-
treatment, would be due to the opening-up of the pore struc-
ture resulting in greater oxygen chemisorption on the surface
of metal particles. However, in spite of the residual oxygen
atoms in the sample after one run, the observed overall weight
loss of the third run was virtually the same as that of the
second run. This seems to have resulted from the decrease in
active metal surface area due to the agglomeration, and Otto

and Shelef (28) found from their X-ray diffraction studies that

43

after an eight—hour heat treatment at 900°C, metal particles
became noticeable in the graphimets, and the diffraction pat-
tern characteristic of the graphimet was gone. All these ob-
servations seem to indicate the difinite increase in the metal
particle size and the structural destruction of the graphimet
during the heat treatment.

In Chapter III, a simple method is developed in the light
of our present experimental results to predict the kinetic
parameters of a moderately heterogeneous system by treating
each phase independently and combining the effect of each phase

together for the net result of the heterogeneous system.

CHAPTER III

THE TEMPERATURE PROGRAMMED DESORPTION/REACTION

APPROACH OF THE GASIFICATIONOF CARBON:

THEORY

44

THE TEMPERATURE PROGRAMMED DESORPTION/REACTION APPROACH

OF THE GASIFICATION OF GRAPHITE: THEORY

A. Theory

We consider the following typical experimental situation

for the theoretical considerations of TPD. The carrier gas,

usually helium, flows through the uniformly grannular catalyst

bed at a constant flow rate, while the temperature of the sys-

tem is linearly raised with time. For the sake of the mathe-

matical simplicity, the following conditions are further as-

sumed:

(l)

(2)

(3)

Intradiffusion within the catalyst bed is neglibile,
and the readsorption of desorbed surface species is
negligible by maintaining a relatively high carrier
gas flow rate.

Gas molecules (oxygen) dissociatively adsorb on the
surface of metal (platinum) particles during the
controlled chemisorption, and do not combine with
carbon until the metal become active at elevated
temperatures. Gas molecules that adsorb on the sur—
face of graphite form surface oxides during the con-
trolled chemisorption at room temperature, T..

The effective surface areas of graphite and metal,
Sc and Sm respectively, are time invariant during
heating, and the desorption kinetics of carbon oxides
from one surface is not affected by that from the

other surface.

45

(4) Homogeniety of graphite and metal surfaces is also
assumed, and the activation energies of desorption
from graphite and metal, Ec and Em respectively,
are independent of coverage.

(5) A first order desorption is assumed.

(6) The metal oxide reduction reaction to form carbon
oxides is not rate determining.

(7) The production of carbon monoxide, CO, is dominant,
and that of carbon dioxide, C02, is neglected.

(8) The Arrhenius rate expression adequately describes
the desorption kinetics, and the frequency factors
assume a constant value at 1013 sec-l.

A schematic picture of the theoretical system is given

in Figure (III-l).

 

 

 

 

 

 

 

He .
HQ ceo '
-————v 4————o-
——"’ c—o
-—.- ____,.
go ——o
00 c" .0 O
o c 0 0 "cf 0‘
O o 0' 0 0
Ftp—JR I 1 I i I \ f Y 1°
graphite layer

 

 

 

Figure III-l Schematic Sketch of TPD System

46

The material balance of CO in the system is given by

dnco ~
—————— = F . - F + rate of CO production
dt co,1n co,out
= - + -
0 CCO Q SRCO (III 1)

Assuming no accumulation of CO in the system,

dnCO _
-_EE_— — 0, SRCO — CCO Q (III-2)

The production of CO originates from two sources, viz,
a) active sites on carbon surface, usually edges, and

b) active sites at the metal-carbon interface.

Hence,
_ c m
SRco ’ Sc Rco + Sm Rco
AC Am
AC exp(-Ec/RT) nCO + Sm Amexp(-Em/RT) nCO

(III-3)

l. TPD of pure carbon surface (or pure metal surface)
Since Sm = 0,
c Ac
SRCO = SC RCO = SC AC exp(—EC/RT) nCO (III-4)

The peak temperatures of TPD spectra are obtained when

the production rate of CO is maximum.

d C _- C EC
-EE_(SC RC0) - SC RC0 -§T§- - AC exp(-EC/RT)

= O (III—5)

47

Therefore, at the peak temperature,

 

c BC BC
2 ln TM - lnP = ——-C-- + 1n( ) , (III-6)
RTM ACR
which is the same form as Equation (I-l). There are two un—
knowns in this equation, viz, Ec and AC. If AC can be assumed
to be approximately 1013 sec-l, EC is uniquely determined from

one experimental TM. On the other hand, EC and AC can be direc-
tly determined by considering the inflection point(s) of TPD

spectra as well.

 

d2 c
When t2 (SC Rco) 0,
EC 2 C
R(TE)2 -;§—— + AC exp(—EC/RTi)
2,
EC c
————-——— — AC exp(-Ec/RTi) (III-7)
R(TE)2

Equations (III-6) and (III-7) can be simultaneously solved for

unique Ec and AC.

2. TPD of Metal Catalyzed Carbon

At the peak temperature, TM ,

d C EC
-———(SRCO) = SC RCO -——7—— - exp(-EC/RTM) AC
dt RTM

m Em
+ Sm RC0 W - exp(-Em/RTM) Am
M

= 0 (III-8)

48

Unlike the single site case, the peak temperatures do not
necessarily relate to the activation energy of desorption of
either carbon or metal. If indeed a first order desorption
exists on both carbon and metal surfaces and the activation
energies of desorption are independent of coverage, the TPD
spectrum of carbon/metal system will be the sum of that of each
separate surface, and the shape of the resulting TPD spectrum
will depend on the relative magnitudes of Em and EC, Am and AC,

and SI“ and Sc-

3. Determination of TPD Spectra of Carbon-Metal System

The rate of CO dissipation in the carbon phase is given by

d c _ AC
:1: SC nco) - - SC AC exp(-EC/RT) “co

= - sC kc exp[’)’c(l - TO/T)]?1:O

(III-9)

£2
5
m
H
m
o2
l

— EC/RTO , and

W
0
ll

AC exp(-’YC)

And the following transformation is further carried out.

v = T/TO , v 2 l
c _ c c c _ c _
uCO — nCO/nco(l) , where nCO(1) - nCO at v — l

49

Then Equation (II-9) reduces to

 

c

duco " k(: 1 C

-—53__ _. #5 exp [‘);(1 _ V )] uco (III-10)
where ”:0 (l) = 1

By the same analogy, the rate of CO dissipation in the

metal phase is given by

 

m
du k l m
dv _ ,3 exp [VIMl v )1 “CO
he m (1) = 1 (III-ll)
W re uCO

Therefore, by the assumptioncfi first order desorption of
CO from both the carbon and the metal phase, the rate of CO

appearance in the gas phase is

dnCO _ . _ 1 Ac
T - n - SC kC exp[’)2(l ' 7)] nco

m
+ Smk m exp [7m (1 - -—) ’nCO (III-12)

If it is assumed that

AC
sc/Sm ~ nCO (1)/nCO (l) I

AT .m

then nCO (1)/nCO (l) = l + Sc/Sm and

AT C
nco (l)/%co

1 + sm/sC

50

Using this assumption, Equation (III—12) can be rewritten

 

as
H = (S /S )(l + S /S )-l k exp P7’(l - -l—fl uC
Sm’figo (l) C m 111 C C C V CO
'1 1 m
+ (l + SC/Sm) km exp [7m(l - 7)] uCO
(III-13)

Equation (III-l3) is the final form desired in the deter-

C

Co and

mination of TPD spectra of carbon-metal system. Once u
Ugo are determined from two first order nonlinear ordinary

differential equations, Equations (111-10) and (III-ll), using

. 0 AT
appropr1ate values for EC, Em, AC, Am, and , n/SmnCO (l)
(Turnover frequency, TOF) can be plotted against v for different

values of Sm/SC (SRATIO ).

These plots are shown in Figure (III-2) through (III—9).
In those figures, EC was assumed to be 260 KJ/mole and 270 KJ/
mole, typical values for uncatalyzed oxidation of graphite, and
Em was varied along with Sm/SC. AC and Am were assumed to be

3 - .
constant at 101 sec 1, andfi was taken to be lOK/min.

THEORETICAL TPD PLOTS

TOF

x 1
sec

0-3
-1

51

 

 

SRATIO =
0.5

 

 

 

 

. \A

2.5 3.0 3:5
T/TO

 

Figure III-2 Theoretical TPD Spectra of
Carbon/Metal

EC = 260 KJ/mole, Em = 230 KJ/mole

TOF

x 10
sec"1

52

 

 

SRATIO
0.5

 

 

 

 

 

l l

 

2.5 3.0 3.5

T/TO

Figure III-3 Theoretical TPD Spectra of
Carbon/Metal

EC = 260 KJ/mole, Em = 250 KJ/mole

53

 

 

 

 

 

 

TOF
x 10-3
sec’1
3.0 -
SRATIO =
0.5
2.0 ’
1.0
1.0 .
0.0 J " ‘ ‘
2.5 3.0 3.5

T/TO

Figure III-4 Theoretical TPD Spectra of Carbon/Metal
EC = 260 KJ/mole, Em = 270 KJ/mole

54

 

 

 

 

 

 

 

 

TOF
x 10"3
sec"1
3.0 .
SRATIO =
2.0 f 0.5
1.0
1.04 (
oo ‘ ‘ L
2.5 3.0 3.5

Figure III-5 Theoretical TPD Spectra of
Carbon/Metal
EC = 260 KJ/mole, Em = 280 KJ/mole

55

 

 

 

 

 

 

TOF 3
x 10-
sec-1
3.0 t
SRATIO =
0.5
2.0 b
5.0
3.0
l O t 1.0
1.0
0.5
,0
,0
0.0 J 1 1
2.5 3.0 3.5 T/TO

Figure III-6 Theoretical TPD Spectra of Carbon/Metal

EC = 260 KJ/mole, Em = 300 KJ/mole

56

 

 

 

 

 

 

 

TOF
x 10-3
sec"1
3.0 ”
SRATIO =
0.5
2.0 '
1.0 b
1.0
0.0 ' 1 A
2.5 3 0 3.5

Figure III-7

Theoretical TPD Spectra of
Carbon/Metal

EC = 270 KJ/mole, Em = 250 KJ/mole

57

 

 

SRATIO
0 . 5

 

 

 

 

3.0 3.5 T/TO

Figure III-8 Theoretical TPD Spectra of Carbon/Metal
EC = 270 KJ/mole, Em = 300 KJ/mole

TOF
x 10'3 sec

58

 

 

''3.

+2.

1.

Figure III-9

SRATIO =
0.5

 

 

 

3.5

 

 

Theoretical TPD Spectra of Carbon/Metal

EC = 270 KJ/mole,

Em

350 KJ/mole

59

C. Discussions on Theoretical TPD Spectra

The surface of the carbon-metal complex is definitely
heterogeneous, consisting of the carbon phase and the metal
phase dispersed along the graphite layers, but the carbon phase
is assumed to have the same characteristics as the pure graph-
ite in the discussion of the theoretical TPD spectra.

For EC = 260 KJ/mole, the peaks of CO spectrum from the
carbon phase are always found at T = 3 TO (Figures III-2, III—5,
and III-6), and for EC = 270 KJ/mole, the corresponding peaks
are at 3.12 To (Figures III-7, III-8, and III-9). This implies
that the temperature difference of about 40°K changes the acti-
vation energy by 10 KJ/mole, and this observation can be used
as a method of estimating the activation energy of pseudo one
phase in a heterogeneous system.

As can be seen in Figures (III-3) and (III—4), a single
peak for CO suggests that the difference between the activation
energies of desorption in the metal and the carbon phase is in
the range of 10 KJ/mole for all ratios of active sites, and for
ratios greater than 3 (Figure III-5) or less than 0.5 (Figure
III-7), the difference can be in the range of 20 KJ/mole.

Although the figures were based on the assumption of the
negligible readsorption which is unavoidable even in a high
vacuum, these differences in activation energies are likely to
remain the same, even if the readsorption may influence the
desorption from both the carbon and the metal phase.

Since all the peaks for the graphimet were at higher tem-

60

peratures than those of the pure graphite in the experimental
results, it can be concluded from the theory that the activa-
tion energy of desorption from the metal phase is greater than
that from the carbon phase.

In Figure (III-5), for the ratios of active sites greater
than 3, a single peak is expected, but one important aspect of
this figure is that for all ratios greater than 3 the peak
temperature is the same at 3.25 To, while the experimental
peaks shifted to higher temperatures with repeated heating.

If it can be assumed that the surfaces are constantly
changing during product desorption and heat treatment, and, so
are the ratios of metal to carbon active sites, then the ex-
perimental observations suggest that the case where the dif-
ference in activation energies is in the range of 20 KJ/mole
as shown in Figure (III—5) is not applicable to our experimental
results. In Figure (III-4), where the activation energy from
the metal phase is 10 KJ/mole greater, the peak temperatures
are seen to shift to higher temperatures as the ratio increases.
If the active sites are changing upon desorption and heating
such that the ratio of metal to carbon active sites increases,
this then would be the case adequate to our results.

As discussed earlier, if indeed the annealing of basal
plane defects causes the decrease in the active area in the car-
bon phase, the decrease in the active area due to metal agglom-
eration has to be less significant in order for the ratio to

increase upon heating and subsequent cooling.

61

Another possible explanation for the increase in the sur-
face area ratio would be due to the opening up of the pore
structure upon continued heating thus, allowing the metal parti-
cles to be more exposed to gas molecules resulting in an increase
in the effective area of the metal. Thus, the total gasifica-
tion rate increases as well.

Nonetheless, the opposite trend is observed in Figure
(III-3) where the peaks shift to higher temperatures as the
surface area ratio decreases. This is not applicable to our
experimental results since in all cases the peak temperatures
are lower than that for graphite. It was shown experimentally
with graphite that the peak shifted to lower temperatures with
repeated heating. The opposite trend observed in the graphimet
sample suggests the following. A shift towards the higher peak
temperature indicate a much larger contribution of the metal.
Hence this further supports our contention that practically
for all cases the ratio, Sm/SC is increased.

The gasification rate can be affected by the catalysts in
two ways, namely, the activation energy and the preexponential
factor. In the theoretical argument presented in this chapter
the rate was expressed in terms of unit area of active sites,
while the preexponential factor was held constant with varying
active sites. Hence, the catalytic effect as viewed from the
rate expression is represented by either a decrease in activa-
tion energy or an increase in the active sites. A typical dif-

ference between the peak temperature of graphite and that of

62

graphimet was experimentally shown to be about 600K (Figures
II-4A and II-8A), and this corresponds to a difference of

15 KJ/mole in the activation energy, which is about 5% of the
reported value for graphite. This increase is well within the
experimental error, or may have been caused by the apparent
lowering of the diffusional barriers as discussed in Chapter II.

If indeed this was the case, the increased gasification
rate observed in the graphimet sample must have been due to the
increase in the active surface area with little or no change
in the activation energy. According to our argument this is
true only if the active sites are changing upon repeated heating
such that the ratio of the metal sites to the carbon sites is
increasing, as depicted in Figure (III-4).

Cases have, in fact, been reported (31) where the activa-
tion energies for the catalyzed and uncatalyzed gasifications
were the same in spite of large differences in the overall
rates. McKee (32) suggested from this observation that the
effect of the catalysis was to increase the preexponential fac-
tor of the rate equation, which would be the expected result
of the rate equation, which would be the expected result of
an increase in the density of reaction sites on the carbon
surface. He also concluded that a real reduction in activation
energy, as distinct from a decrease due to mass transport limi-
tations, would imply that the catalyst effectively increases
the rate of the limiting step in the reaction sequence.

Many of these experimental observation on the apparent

63

activation energy and the preexponential factor were explained
in terms of the compensation effect (33), but no satisfactory
answer concerning the influences of catalysts on the kinectic
parameters of the uncatalyzed reaction has been given since
several important variables such as mass— and heat transfer
limitations and continuous changes on the gasified surface
were neglected.

The simple theory presented in this chapter seems to work
fairly well in estimating the relative magnitudes of activa-
tion energies and the changes in the active sites of a moder-
ately heterogeneous system during gasification. One improve-
ment that can be readily made on this theory would be to incor-
porate some kind of functional dependency of the activation
energy on coverage, in addition to the already added hetero-
geniety by treating each phase separately. Amenomiya and
Cvetanovic (2) assumed that the composite activation energy was
a simple linear function of surface coverage with an upper and
lower limit, and obtained theoretical TPD spectra with a much
broader peak than that of constant activation energy.

With these kinds of further improvements on the hetero-
geniety incorporated into the present theory, an interpretation
of good experimental results will bear a lot more meaningful

details to help us for the better understanding of the mechanism.

CHAPTER IV

THE CHANNEL PROPAGATION RATE OF PLATINUM ON A CARBON SURFACE

A MODEL AND STEM STUDIES

64

BACKGROUND

Over the past decades, a considerable amount of informa-
tion has been collected on the catalytic influence of metals
on the gasification of carbon under various conditions. It
is not until recently that some details of the catalytic acti-
vity of these metals have been investigated. Optical micro-
scopy and most importantly, Controlled Atmosphere Electron
Microscopy (CAEM) has allowed a detailed picture of the acti-
vity of metal catalyst undergoing reactions. The catalytic
influence of platinum (27, 34, 35, 36, 37) on the oxidation
of graphite has been reported previously. Presland and Hedley
(35) used electron microscopy to investigage the Pt/graphite-
oxygen reaction at 600-1,000°C, and found that the metal parti-
cles were only active as long as they contacted edges or steps
on the graphite surface. Some particles produced pits while
others generated channels across the surface. L'Homme et a1.
(36) investigated the same reaction and suggested that oxygen
dissociated on the platinum surface and the atoms produced mi-
grated over the catalyst particle to the carbon. Just recently,
Baker et a1. (16) investigated the catalytic oxidation of graph-
ite by platinum employing CAEM. In their study, they found
that at 500°C, the average particle size grew and in some in-
stances were penetrating the graphite basal plane to produce
pits. The pits were subsequently expanded by edge recession

due to the uncatalyzed attack and thereby producing hexagonal

65

holes. However, upon continued oxidation, the pit increased
in depth and the uppermost graphite layers become progressively
more circular in shape.
At 735.C, another mode of catalytic attack was occurring.
Those particles which had previously created pits initiated
an action parallel to the basal plane to form channels, with
the particle at the head of the channel. They have also obser-
ved that channels were also being produced by other particles
which had contacted edges or steps on the graphite surface.
With respect to the particle mobility, channels propagated by
small particles (up to 25 nm diameter), were relatively
straight lines, making angles of 600and 120.to one another,
and were usually oriented parallel to the (1120) directions.
Particles greater than 100 nm diameter propagated irregular
channels, which possessed hexagonal facets at the graphite-
catalyst interface. Here the facets are oriented parallel to
the (1150) directions.
some interesting behaviour of the particles were also
observed and are common to all channeling particles:
(1) The width of the channel was dictated by that of the
catalyst particle responsible for its propagation,
and often changed as the particle spread or contracted
at the graphite-metal surface.
(2) In some instance, the particles break and the fragmen-
tary particles continued to channel but at a faster

linear rates than that of the progenitor. Conversely,

66

active particles collided with inactive stationary
particles and coalescence occurred. At this instance,
the channel propagation was arrested for a brief time,
and finally the new particle accelerated to reach a
new but lower constant channel propagation rate.

(3) Loss of activity occurred if the particle lost contact
with an edge on the graphite surface. At higher tem-
peratures, such particles often contained sufficient
kinetic energy for motion to locate a fresh edge or
step and continued to propagate channels.

(4) A spherical form is assumed by particles which tempor-
arily lost their activity (i.e., mobility). In con-
trast, inactive particles, which remained stationary,
were irregularly shaped. At 8500C, there was general
particle mobility and all particles became spherical
in form.

A striking observation was found by Baker et a1. (16)

in realtion to channel propagation rate and particle size.
Their quantitative kinetic analysis showed that for a given
temperature, the linear rate of channel propagation was in-
versely proportional to both the catalyst particle size and
the channel depth. It is the purpose of this section to prove

some of this effects by a model.

67

SCANNING-TRANSMISSION ELECTRON MICROSCOPY OF
UNREACTED AND REACTED PLATINUM GRAPHIMET

Here, we briefly present some of the electron micro-
scopies of both unreacted and reacted platinum graphimet.
Since Baker's experiment used a different system, viz, plati-
num evaporated on a single crystal graphite (from Ticonderoga)
and platinum (from drying of chloroplatinic acid) on a single
crystal graphite, we investigated the behaviour of platinum
particles for graphimet, and then determine if there is suita—
ble agreement with Baker's results.

Figures (IV-l) to (IV-4) show a scanning transmission
spectrocopy of an unreacted platinum graphimet. All the elec-
tron micrographs use the bright field emission from a STEM
( DtflelIEBSOllanGihmmrmmxms,Inc.).

Figure (IV-l) is a 100,000 magnification (1,000 A/cm) of
a graphimet flake situated on a holey film on a copper grid.
Clearly, the graphimet we obtained from Alfa Corporation is
not purely an intercalated one but instead, represents a combi-
nation of very highly dispersed platinum and large clusters
of smaller groups of platinum. The platinum shown as highly
dispersed and clustered particles are verified by using an
X-ray dispersive analyzer. Some dark points within the speci-
men, represented copper and this was verified by having a very
high correlation from X-ray analysis for copper as shown

in Figure (IV-5). For instance, in Figure (IV-l), sections

68

D and E represent copper. This is not surprising because the
background used was a copper grid,and it was very difficult
either to prevent during the sample preparation the tearing of
the plastic holey film or placing snuggly miniscales of graphi-
met flakes within the boundaries of the copper grid. Further-
more, the hybridization of transmission electron microscopy
with scanning electron microscopy allows to show the depth
profile of the specimen.

Hence, a slight opening of the graphite flakes can be
visibly detected by the scanning mode.

Our observations on the unreacted graphimet support the
findings of Smith et a1. (38). Their observations reveals
that the intercalated element was present primarily as small
islands of either metal or metal oxide on the surfaces of their
graphite flakes which appeared to have been separated as a
consequence of the manufacturing procedure. In order to obtain
a yet better resolution of the particles, we increased the
magnification of a certain section in Figure (IV—l), section A.

Figure (IV-2) is an expanded view of section A with a
500,000 magnification (500 A/cm). Here, we notice the clusters
of particles having sizes in the order of about 20 A. Another
striking feature of this photograph is that fringes can be
visibly seen. These lattice fringes arise from openings between
graphite planes which contain some clusters of platinum (Section
G, Figure (IV-4)). This prompted us to further increase the

magnification and examine the platinum intercalates. Figures

69

(IV-3) and (IV-4) show a 106 (SA/cm) and 2 x 106 (ZA/cm) magni-
fication of section B. Note the lattice fringes have become
more visible. All dark and somewhat round spots were identi-
fied as platinum, using again the X-ray dispersive analyzer.
For all sections of the platinum graphimet investigated, all
showed small and large clusters of platinum.

There are two findings that revealed by these micrographs
which have not yet reported in the literature. First, the
platinum graphimet, aside from not being an intercalated speci-
men, contains small amounts of impurity. For the sample used,
X-ray dispersive analysis detects low levels of chlorine and
iron. Although it is not reported here, other sections in the
specimen reveal the presence of small amounts of silicon. In
all TPR and TPD experiments it was assumed that at these low
levels, the impurities do not impart any other catalytic acti-
vity. This is rightly so for the following reasons. Silicon
is catalytically inactive and probably will not form an unre-
active Pt-Si phase. Chlorine can be removed during the suc-
cessive pretreatment with oxygen. Although iron processes a
catalytic activity in the steam/H2 gasification of graphite,

a pre-exposure with oxygen and conducting a series of similar
pretreatments will enable iron to form a more stable but un-
reactive iron oxide. Second, in contrast to the findings of
Fisher et al. (38), it was observed that the islands of plati-
num which are very small are not less numerous. However, other

micrographs confirm Fisher et al.'s findings that there is a

70

tendency of the particles to cluster along crevices between
two overlapping flakes (Section F in Figure (IV-1)), or at
surface steps, although there did not appear to be any pre-
ferred orientation of the platinum lattice planes relative to
that of the graphite.

The electron micrographs for the reacted platinum graphi-
met do not resemble those of the unreacted. Here, the reacted
platinum graphimet is referred to as the sample which had
undergone the temperature programmed desorption/reaction as
described in the Chapter II. In all the electron microscopies
of the platinum graphimet flakes, it is observed that the
platinum particles tend to be more spherical as was also ob-
served by Baker et a1. (16). Furthermore, the particles are
more organized and are not highly dispersed into small islands
as that found for the unreacted sample. It can thus be spec-
ulated that during the few minutes of ramping of the tempera—
ture (after pre-exposure to oxygen), the highly dispersed
platinum islands become mobile and agglomerate (or sinter) to
form larger particles. In the TPD/TPR set-up, the sintering
occurred in the absence of an external oxygen flow, in contrast
to the experiment of Baker et al. (16), where l torr of oxygen
was flowed continusouly over the Pt-graphite. Figure (IV-6)
is a STEM of a reacted platinum graphimet with a 50,000 magni-
fication (2,000 A/cm). A number of possible catalytic acti-
vities can be conjectured from this electron microsgraph.

Clearly, particle channeling is one such catalytic activity.

71

A 8}. see; c “SHEETS @8085 no EB TB 685E
O

 

A EU\< ocm v ugsmmumlum .pwuomohas m0 Em NI>H mafia
0

 

73

EU? om o aflhzmmuanum 6308.5: mo gm
0

MIPH 053m

 

74

A no} om V ugnmmnouum Umuomouc: mo Em
0

VIE“ gem

 

m-IICO()

75

 

 

 

 

LT-lflfl SECS UNREACTBD CARBON 2
4»
1500-0
S
f P
( 1' -
19?0-( .
1' '
0
‘P P
. T
V . '
SCH-0 .
.L
.P.
P. ‘
0 O . C ° .0
. . O ‘1 .
C "
4} .’:. L o I 0
'.° ' 3:. (3 ' '2 ° ° ' ' I"
P .I’. ‘4‘ .‘ . .‘ o o 2 T F
. o o ' . . A
0 it ’aTflT‘Mfl‘?hh~nd-qfl-r’ ‘9 7 ‘uq.u: 1» f-
- l I I I l ‘ I I l I M W
0.000 5.000 10.090 1 CT I 15.000

ENERGY (KEV)

Figure lV-S x-ray dispersive analysis of Pt-graphimet

76

A 6}. o8
0

L

m o

ugnmmhwlum moouomou mo "Em

w

13 953m

 

77

A 16>“ 084 o ugslassoufi @888 no 5.8 TB ESE

 

78

A 8).... com o 6.6528813 8888 mo 5.8.

TB 686E

 

79

A62 OOHV
O

ugcamnmrum @30me mo «Em

m

(B 059E

 

80

Figures (IV-7) and (IV-8) are a 100,000 and 200,000 magnifi—
cations respectively of the same platinum graphimet flake.

The narrow channel runs from the middle upper left to the lower
left portion ofthe picture. X-ray dispersive analysis reveals
that the channels do not contain any platinum except near the
end points. The Figure (IV-10) illustrates more clearly the
situation. Platinum particles within the channel have sizes
ranging from about 100 to 150 A. To the right of the narrow
channel is another channel. However, this channel is wider
but shorter and contains a larger platinum particle (~375 A).
This observation was also made by Baker et al (16), where
larger particles tended to have slower propagation rate than

smaller particles.

 

‘

 

channel

1

 

 

 

Figure IV-lO Schematic Illustration of Various
Channelling Activities in Figure (IV-7)

81

Figure (IV-9) is a one million magnification (100 A/cm)
of the same reacted platinum graphimet. The large dark spot
at the upper right corner is platinum (as determined from the
X-ray dispersive analyzer) having a particle length of about
400-450 A. This electron micrograph shows that the particle
did not exhibit a channeling activity. It is also observed
the graphite lattices are intact and unaffected by the small
gasification activity (because of the low level of oxygen
exposure). Although there was no STEM of a reacted graphite
such that no direct comparison could be made, it can be sum-
marized however from these electron micrographs that most of
the catalytic activity occur within the vicinity of the plati-
num particles.

Another catalytic activity is the pitting mode. There
seem to be no clear distinction from these electron micro-
graphs of such activity and will require specifically a scan-
ning electron microscopy. However, some particles of the
electron micrographs reveal somewhat that pitting action might
have occurred. The lower left section of Figure (IV-6) pro-
bably depicts a relatively good STEM resolution of the pitting
mode.

Six conclusive statements can be made in regards to these
studies. First, there is a resemblance of catalytic activities
between Pt/graphite and platinum graphimet. Hence, it can be
hypothesized that such activities could be occuring for other

Pt/carbon system, e.g. Pt/char. Second, the TPD/TPR set-up

82

as describe:U1Chapter II can be a very viable tool for de-
termining intrinsic kinetic rates of gasification since turn-
over rates can be calculated on an almost constant surface
metal area. This is equivalent to studying the gasfication
kinetics in a differential mode where low levels of oxygen
exposure lead to low conversions. Furthermore, for low con-
version, diffusion controlled reaction is almost eliminated,
hence we are confident that the activation energy obtained is
not apparent but true activation energy. Although this was
not discussed earlier, an electron microscopy of the platinum
graphimet at various temperature conditions in the TPD/TPR
experiment should reveal an almost similar morphology (size
and shape) of the platinum particles. This was not executed
in this study because of a certain difficulty of the imple-
mentation. It is conjectured here that particles first re-
arrange at lower temperatures even at temperature below any
incipient gasification activity. Hence, once a definite
morphology is established it would have remained as such until
the end of the TPD/TPR run. From this, the assumption of
constant metal surface area during that part of the catalytic
activity is not far flung. Third, the catalytic activity
occurs within the particle's vicinity, thus requiring a good
contact between metal and support (carbon). Fourth, platinum
can sinter in small amounts of oxygen (from desorption), car-
bon monoxide and carbon dioxide as reacted by the TPD/TPR ex-

periments and electron microscopy. Hence, metal-support

83

interactions are not probably strong. However, it is sur-
prising that platinum exhibits good catalytic effectiveness.
And furthermore, it was not mentioned earlier that there

should be a good contact between metal and support for any
catalytic activity to occur. Here, the excellent contact
between metal and support is not a measure of SMSI (Strong
Metal Support Interactions). It is viewed here that the con-
tact is a result of the catalyst performing a depolymeriza-
tion or carbon-carbon bond breaking. Boudart and Holstein

(14) support such idea. They are rightly so because the hydro-
genolysis of carbon-carbon bonds is a limiting process in the
hydrogenation of carbon to methane. We speculate that from

our data based on TPD/TPR experiments, platinum functions as

an agent for carbon-carbon bond breaking. The following argues
such validity:

First, assuming (1) a first order rate for both catalyzed
and uncatalyzed reaction and (2) the reduction of the metal
oxide largely byaicarbidic carbon and desorption of surface
carbon oxides are not rate limiting, and it is found that the
activation energies of the uncatalyzed and catalyzed rates are
just slightly different from each other. This implies that
the carbon-carbon bond breaking is rate limiting. From the
TPD/TPR experiments, the activation energies are 310 KJ/mole
and 300 KJ/mole for the catalyzed and uncatalyzed rates,
respectively. This also implies here that platinum does not

alter the reaction pathway of the uncatalyzed rate. The slight

84

lowering of the activation energy is usually interpreted as

an enhancement of the carbon-carbon bond breaking. Second,
from the TPD/TPR spectrum, the incipient gasification (ex-
pressed in terms of any surface oxide evolution, mainly car-
bon dioxide) occurs at a lower temperature (SOD-510°C) for
platinum-graphimet than graphite (600-620°C). In the conduct
of the TPD/TPR experiments, one observation is worth noting.
The differences in the temperature at incipient gasification
between catalyzed and uncatalyzed rates could be hardly de-
tected from the weight loss data. It would require a very
sensitive electrobalance to detect a weight loss in the order
of 10-9 mole. This was circumvented here by measuring the gas
evolution using a sensitive gas Chromatograph. Hence, our
TPD/TPR experiments reveal that gasification could start at a
much lower temperature than what is usually reported in the
literature by weight loss data. Thus, platinum acts as an
agent for implementing the carbonacarbon bond breaking leading
to the formation of a carbon-platinum bond with subsequent
reaction of the carbidic carbon with a thin layer of metastable
platinum oxide. Here the metastable platinum oxide could con-
sist of a weakly bound oxygen, as envisioned by Holstein and
Boudart (15) and perhaps not forming a separate phase in con—
trast to the suggestions by Baker and Coworkers (17, 18).

This leads us to the fifth conclusion. We concur with Baker's
suggestion that there is probably a thin layer of metal oxide

which could be metastable and becoming non-thermodynamically

85

or non-stoichiometrically unstable at higher temperatures
because of its subsequent reactions with the carbidic carbon.
This is supported from the TPD/TPR experiments where both
graphite and platinum graphimet were preexposed to equal
amounts of oxygen. A.weakly boundatomic oxygen on the platinum
surface arise from a dissociative chemisorptioncfifoxygen, even
at lower temperatures. We have made no verification as to
whether adatomic oxygen reacts or diffuses within the reduced
metal matrix at lower temperatures. The former is very possi-
ble on iron surfaces. For longer oxygen exposures diffusion
within the platinum crystals is not a remote possibility. At
the outset of the temperature ramping, some weakly bound oxygen
is desorbed and carried away by the inert gas and the rest of
the bound oxygen partake to form the metal oxide during the
rearrangement of the particles (e.g. redispersion and sintering)
and at which time there is yet no gasification activity.

Hence, preceeding any gasification activity, the particles
could have two phases which are not necessarily distinct and
for which the oxide phase is not necessarily a thermodynamically
stable metal oxide. Furthermore, at roughly the same initial
weights for both graphite and platinum graphimet, the latter
has a larger initial oxygen uptake as clearly demonstrated by
the total evolution of the surface carbon oxides (CO and C02)
in the TPD/TPR spectrum (Figure II-4B and II-8B). This sug-
gests that any dissociated oxygen at exposure temperature is

nearly maintained intact (except perhaps for small desorption)

86

until its reaction with a carbidic carbon at elevated temper-
atures. Moreover, it is evident that in the absence of an
external flow of oxygen, a huge level of CO is detected at
temperatures greater than 700°C with a lapse time of nearby

80 minutes before the first detection of carbon monoxide.

This suggests that a source of oxygen has been kept for the
CO production, and it can be conjectured that since there is
no other source of oxygen, such oxygen source could have been
supplied from an unstable metal oxide layer. The occurrence
of adatomic desorption of oxygen is probably a remote possi-
bility and if it occurs at all it will be at a minimum level
and they have to be desorbed at lower temperatures. However,
with the temperature ramped such possibility is virtually nil.
Moreover, on the basis of thermodynamic argument, it is almost
impossible to reduce Pt-O to Pt + 802 by a simple heat treat-
ment. Hence, once an oxide is formed at relatively lower
temperatures it will remain as such. We believe our TPR/TPD
results strongly suggest the existence of a metal oxide layer
(be it stable or unstable) and is especially supported by the
fact that there is no continuous flow of external oxygen in
the TPD/TPR experiments. We are not discounting the existence
of a weakly bound oxygen. This is probably present in a con-
tinuous flow of oxygen. However, we cannot totally disregard
the existence of a thin layer of metal oxide as our TPD/TPR
experiments strongly suggest. In the next section of this

Chapter, we have set forth a mathematical model which includes

87

the existence of a thin layer of metal oxide and show that
this model supports our theory and that of Baker for it matches
those of Baker's (16) data on channelling rate.

Finally, we think that an oxygen transfer mechanism is
also operating for the platinum catalysis. This is supported
by the following. The activation energies of both catalyzed
and uncatalyzed rates are not distinctly different and yet the
reaction rate is more than doubled in the former. The enhance-
ment in the catalyzed rate can be explained by a compensation
effect (39, 40, 41) where the pre-exponential factor has values
much greater for the catalyzed than the uncatalyzed rate. The
pre-exponential factor is related to the active sites density.
Hence, it is clear that platinum creates and furnishes more
active sites density. Otto and co-workers (31) found the same
conclusion for the nickel-catalyzed steam gasification of char
and pure graphite. Boudart and Holstein (15) seem to support
the oxygen transfer mechanism for the oxidation of carbon by
transition metals.

Now, Otto and co-workers (31) interpreted the increase
in active sites density as due to oxygen transfer rather than
by electron transfer. For the latter to occur, they suggest
that there should be a substantial decrease in the activation
energy for platinum-graphimet. However, an ease in the break-
ing of the carbon-carbon bond would imply an increase in the
rate of formation of active carbon, thus increasing the number

of active sites density which is a compensation effect. Here,

88

we therefore view the two mechanism, viz, oxygen and electron
transfers as totally distinct and so it is possible that both
mechanisms are occuring in any gasification of carbon. That
the latter is rightly so can be viewed in terms of the viscous
metal oxide layer theory we hypothesized earlier. To decipher
which mechanism is rate limiting seems to be not clear from
our experiments and would require a more detailed kinetic

study. One suggestion is to use tracer techniques.

 

cata\lyst

 

 

oxide

 

 

 

 

 

 

 

 

ra hite
9 P FLA

(b)

 

 

 

Figure IV-ll Simplified Sketch of Particle
Channelling Activity - Model
(a: a general View, b: two dimensional view)

89

A MATHEMATICAL MODEL
OF THE CATALYST PARTICLE PROPAGATION RATE
DURING AN ACTIVE GASIFICATION OF GRAPHITE

Much has been speculated about the catalyst behaviour
during gasification. Crucial to all of this is the existence
of a thin viscous layer of metal oxide. Boudart and Holstein
(15) hypothesized that since noble metals, which are only
present in the reduced state, are able to catalyze the C-02
reaction is an evidence that two phases are not needed. Baker
and co-workers (17) on the other hand suggests its existence.
By measuring the channel propagation rate through the use of
CAEM, they have found an underlying distinction between catal-
yzed oxidation and catalyzed hydrogenation. In the former,
the rate at which particles propagate or create channels is
inversely proportional to some dependency in the particle size
(usually square root of particle width) while for the latter,
the linear rate of channel propagation rate is proportional to
some function of particle size (square of particle diameter
for the nickel-catalyzed hydrogenation of graphite). They have
envisioned that in a strong oxidizing environment (as in the
C-02 reactionh after the scission of the carbon-carbon bond by
platinum at the graphite-catalyst interface this carbon is
dissolved in the particle and diffuses through a viscous outer
layer of the particle. At the cooler part of the particle,
carbon is converted to either or both carbon dioxide and carbon

monoxide by reaction with atomic or molecular oxygen, thereby

90

creating a carbon concentration gradient within the particle.
While in reality this process is seemingly quite complex,
consisting of a number of steps, and any of which could be
rate determining, we have thought our model which probably
could explain Baker's hypothesis.

Here, we have considered a single catalyst particle
already creating a channel so that we render devoid the ini-
tial transients such as wetting, spreading and agglomeration
of particles. Although Baker and co-workers contend a non-
isothermal particle, it is regarded here as a first approxi-
mation that the particle is isothermal as well as the support
(i.e. carbon). The thin viscous (outer) oxide laye is well-
established and provides a concentration gradient for carbon
within the reduced metal matrix. Carbon diffuses from the
catalyst-graphite interface to the top surface. A schematic
representation of the process is given in Figure (IV-10).

A two-dimensional view of the carbon diffusion is assumed
since the particle is moving along the channel's path, i.e.,
lateral diffusion is small compared to diffusion from the
bottom and front. There is no diffusion and reaction of car-
bon behind the particle. It is assumed that the diffusion
coefficients are concentration independent. The thickness

of the viscous oxide layer is small and remains constant
throughout the steady-state motion of the particle. There is
a limiting rate of transport of carbon from the "bulk" metal

to the external surface of the particle because of the presence

91

of a thin viscous oxide layer. This rate is proportional to
the difference in concentration of carbon at the metal-viscous
metal oxide interface and at the surface. The proportionality
"constant" is modeled by using a film theory where such con-
stant is determined by the diffusion coefficient of carbon
within the metal oxide and the thickness,6 . The C-OZ reac—
tion only occurs at the top surface, and when oxygen is in
excess, it is assumed that the rate is first order in surface
carbon concentration. Since the concentration of carbon is
distributed along the horizontal, its rate of consumption on
the surface is not uniformly distributed so that both carbon
surface concentration gradient and the mobile viscous nature
of the oxide layer could impose a surface diffusion of carbon
on the surface. Hence, for the metal matrix, the carbon bal-

ance equation in dimensionless form is

 

 

31;: a‘é‘ _
a £1 + a; O (IV-l)
a (0:8) = O (IV-2a)
A
39 - o (IV-2b)
af 5‘"
.23?— = 0 (IV-2c)
a
§= 0

 

 

I~
as A s ._
-32. “MI-E ]_¢£=O (Ix/2d)
2.;=: g... 4

, where C _._ C/Co

’s‘. =I-9.

S S
9.‘-—<=/Co
Co = carbon concentration at the metal-carbon

interface
C5 = surface carbon concentration

5 .
C5'— reference carbon concentrat1on on the surface

4*?
fl

‘X/L , g = Ia/L ( see Figure IV—ll)

92 =' Khl'//[)c"'
Kc

DC ‘MO

6

diffusion coefficient of carbon atoms

mass transfer coefficient =

Dem

in the metal matrix
Demo: diffusion coefficient of carbon atoms

in the metal oxide layer

4 = (2’ (hX—E—i—X—E)

1»
k = a thermodynamic constant relating the surface

concentration to the volume concentration in

the metal oxide layer
L_ = particle size. Here it is assumed, as a

first approximation, the particle is cubic.

The surface carbon balance equation in a dimension-

less form is given by

 

 

 

A
1
4.“? __ 962‘35 ._ £92. =~<3
‘52 ~ 3 C (;=:
5 ._
9 (ea " A
(19.5 __ (IV-4b)
d Eta " C)
, where $1. _ kl}
’ Des
k, = first order reaction rate constant

[ls = surface diffusion coefficient on a

viscous metal oxide layer

Dan L Co
0:50?

JO~
I

 

c‘m

(IV-3)

(IV-4a)

The solution to this set of ordinary-partial differ-

ential equations is discussed in detail in Appendix.

94

Hence, the concentration profile for the carbon

atoms in the metal matrix is given by

s. cl~2Z[AEo C: —'] 5‘">‘nn€‘05h>‘€
nso [abs—TL — Eff-fall- ECHO ’0‘" 55 «(An -7( “as“

 

(IV-5)
_ In H
, where XII- _2— 1r
And the surface carbon concentration is given by
An SlNXflE
=A[ coshég - 111mg smhég] - SD Z W
"80
(IV-6)

x.(2/\¢.-2¢2)
€472 ~54“): wean.

~
, where A
H

The channel propagation rate is given by

-<V> Co - average flux of carbon at the carbon-

metal interface
L
I
if Ne 4‘:
o

, where “L'= flux of carbon atoms

 

1:0

6° k -l
k I.2 ML __ X35
Eks Ekmdhnh)U\ Charo

 

(IV-7)

In order to determine the dependency of propagation
rate on the particle size, we employed the following phys-

ical property data.

k

= 1013 exp(-80/l.987xlO-l3

x1073)
= 5.06 x 10-4 sec-1
Here, E = 80 Kcal/mole at 800°C was based on the value
obtained by Otto and Shelf (28), and the preexponential fac-

tor was taken as 1013 sec"1

, a typical value for k0.
The surface diffusivity of carbon atoms, Dcs' was taken

from the equation,

DCS — 4.7 x 10'2 exp(-21x103/RT)

2.5 X 10‘6 cmZ/sec,

experimentally obtained by Polak (42) for nitrogen on the
(110) plane of tungsten. Due to the lack of the data on the
diffusivity of carbon atoms through the metal oxide layer,
Dc-mo was assumed to be the same as Dcsr 2.5 X 10"6 cmZ/sec.
The diffusivity of carbon atoms within the metal matrix,
Dcm = 2 X 10'7 cmZ/sec, was obtained from the high-temperature

vacuum metallographic studies of carbon atom diffusion in

metals by L'nyanoi (43).

96

The solubility of carbon in platinum was assumed to
be 10 atom %, based on the data by Ershov (44). The con-
centration of carbon atoms at the carbon-metal interface,
CO, was calculated from this solubility data using the

density of platinum of 21.4 g/ml.

Volume of 100 platinum atoms

(100 atoms) (195.09 g/6.02 x 1023 atoms)
21.4 g/ml

 

1.51 x 10‘21 cm3

 

Hence,
C 10 atoms/6.02 X 1023 atoms/mole
O 1.51 x 10-21 cm3
= 1.1 x 10'2 moles/cm3

Assuming the effective thickness of oxide layer to be
9
20 A and knowing the density of carbon atoms of 2.26 g/cm3,
C8 = 4.0 X 10'8 mole/cmz. j\_= CO/C: (at x=0) = 1.0, since
carbon atoms are likely to be supplied through the metal

matrix to the oxide surface efficiently so as to maintain

its saturation level at X = 0.

A plot of particle velocity ( or channel propagation
rate) versus particle width is shown in Figure (IV-12). For
a fixed k, it is clear that there is an inverse relation

between particle velocity and particle width. There is a

97

~
striking match between this model for k = 2.7501 x 105/cm and
Baker's data, as shown in Figure (IV—12), both qualitatively
and quatitatively. Aithough their data is for palladium, our

model seems to suggest the existence of a thin oxide layer on

a metal particle.

98

Channel Propagation Rate
(nm)

 

90 '

80 -

60 —

50 -

30 -

 

Baker's

  

20 -

.\_~
‘

~

 

 

0 l 1 1 1 1 1
10 20 30 40 50 60 70 80

Particle width (nm)

 

Figure 1v-12 (E1 = 2.75005 x 105, 1:2 = 2.75007 x 105,
E3 = 2.75010 x 105, E4 = 2.75011 x 105)

99

Channel Propagation Rate
(cm/sec)

0.5

0.1

 

o—J

 

’L l l I
10'4 104 105 106
’1?

Figure IV-l3 Dependency of Propagation Rate
on k

 

100

CLOSURES

The kinetics of the oxidation and steam gasification
of graphite and platinum graphimet has been presented. The
kinetic studies employ temperature programmed desorption/
reaction. A scanning transmission electron microscopy of
both unreacted and reacted platinum graphimet reveals that
the behavior and morphologies are similar to those found
by Controlled Atmosphere Electron Microscopy. A corre-
lation using a mathematical model of the temperature pro-
grammed desorption/reaction experiments show that platinum
increases the density of active sites. This is attributed
to the scission of carbon—carbon bonds by platinum. The
lowering of the catalyzed oxidation temperature is explained
by the scission function of platinum. For platinum, there
is some possibility that both electron and oxygen transfer
mechanism could be Operating for the catalyzed reaction.
The latter is suggested by a match between Baker's Control-
led Atmosphere Electron Microscopy studies and a simple
mathematical model using the assumptions of surface dif-

fusion and viscous oxide layer.

APPENDICES

Appendix A. Computer Program and Sample

Calculations

4.n+587

Fm

=1

N”

70/175

rRQhRAV TH)

UUTDUT)

INPUTgTfiDEu

91HKK£)91LK9K
XENDQHOECQEMcHRBqPFyGAMMACquMHAVQ

PROGRAM TPD(INPUT~OUTDUT0TAPES

lWTLSLK:
RLAL =

 

1

EXTERNAL

scab
0.5

v-G
OMIIDOIHL
Kira—1N U

Zx>>hltz

 

 

.- _1 f."

0AA l CILAJ

...
lll~lelth-

 

101

(-IQI‘ILIqN‘TAL I

(
((‘10’fir,

A””AM)

 

 

Cl‘bO‘JUbWD

 

 

A
0.
LL.
H
o
X
3
o
X
3
H
o
X
U
D
2
H I
o ' u
M 4
Lu 2'
O- I
H d
I '3
0 v.
I: 0 (if
h- 0 xi—
1...! r0 l-J'U‘
Z *2
O C) (JO
.1 I— ”—0
D m¢
»- o z~
o 9 OC
0 UH
Z as ur—
hJ C) Ad
X I 00'
o 0 H0
). . U—\
0 Lu «I o
I .12) 71"“
9C) 0 0 002+
xoall‘i .
hfl'NAA 0,—0
fl v-uthu
ozo>-4-\
~10"- O>>-\ 0
pm 0 04‘
'44

(UHHI-fl-fl— l
H‘V

KII'Y‘CI
9Y(1)9Y(2)910F

uunnnuvsic-Lcl
(6960)ECQEMQX

 

0
MC

rx

ITE

NTIN
Tn

I’K
HR
C0

0

“1

100

if.
2

uL"

 

59b0)ECvEM9X9Y(1)9Y(2)9TOF

(
(
0
P
7
0
h”

xO'
b>F3<°LLn£LlLJ
V‘.

 

a .
«a? w.

 

as

102

 

 

 

 

 

 

 

 

czu m
zmahmm
xaazozuomoazv> 44mm
21 zuouhzr
.ga > x.z.o:ul m:«»:oxmsm H
aura; mwfi\¢~ prod bzmpaommom
arm
, DH n. U .u...
,n u Hxn»
_mg um Had»
.1u u: mzou
4m uu mzou
$7." .IIPHII
\o no zzqo
m +b Huuo
m. o - o . c H my:
.aomncmmEM1
Viaaoowwnum w
xh7mzou.0hmzovu_777 7 7H
sqzzqosu<zz<o.ud mmxozwouuox..2.wz~ma>..2yfi ,4,4<a¢
7 .. ..........Z KMQJPRH I... ---! Illnl
nquxd> »ux.z.zcm L2Hh30mw3m M
+r.¢ zhu Hnwnc whaxab nu bxhhacocbm

MB

V

1
I
f : 1
l

  
    
    
   
  
   

Q.”

Ut~

; .
. , g
z i ' i I
com ¢3QDC2L1L CoL‘IQL oopcmu-aa-o C'?‘ non—u...)
bog: (“moot . :r-c-sc mmpc.‘ apnoea-r :3.) 13'.

man"- mwcsop L c opooczpcm. c»puc3r:)r«m::ff
uc \_ (...-7. Dog: Ls”. bc: .2: Lo.-.-c c pc;c-. m. - .

' to
no I Lr-b- 000 u Ha-u" coca
..fm7‘7- .UJWWQ‘3‘3N can:
woo moocޢ~ meow
H—w mac»): 5;. (”vocab

maoooo)ooobooopOQOOQQQOQQG HHH #mmmmmm ODD

locumcfi'} -C‘I"“'C C‘C C» CDC!) CVC)€'1 DO.“ HWC‘ JODC‘.
k‘OchOC‘ c 06- 00:3 .-0c c cm: 00:» oc‘cl cacao »
>QC—cacaoo .c (:3: 9:30 cc r. 06 c ooc‘ 3c LJC coo »

0.0.0.0000...
......)

-m-~

 

a: 000.2000 coo o-aomlecoc emu
c: COOL-moo coo $033
:30 OCCi'Cfifit.) DOD C-‘J‘J\
5.60 9005000 9:30 U‘J‘U‘
Zoo OOEtC QC) 03.10 ‘0 7‘3‘
one: 604:: can cacao amp
(.2630 (Danube-:33
0:3 .Doobcoo can ‘Ofla‘xo.
......OO‘OIOOciooo.

«4H ‘ Ho—cHHv-ov-a! Hv-CH

om cacmcmw «aw on“
C301 000430030
Do: 000(30ng

UOC) 00013::va
ZQQ ”306.30 Jr‘vfht‘ m;

 
 

coo coopamc ~0‘U‘O‘3 0333* “N3.“-
uxm :3 0001).) 390‘ 030303 33:30“ .13 '3: ..
ONT-0‘ 3‘0‘0" 7*3‘Cl‘

  

 

ochoo-sbsmc
0.3.0'000000

0.0.0.00
HHHHHv-dH-H

MCDGJU‘dDU"
emmmwFN,
. O . O O O . .
NNNWNNN

ozrro Dmmbmom
II-SH 103.0%th 420$
\ooOOIVOOOOOOO-O'OOOO.
FHH(HHHH1HHv-OH

 

 

 

 

A

5 I !

vac) 130019093 D Doc: DOO'
ogoo0,0oooooO-o 000.00...

.133 Doom-{430.3 _ .1, '33:) 30053

(DO QOOVD’TVDC) :2 ODE) DJC‘I‘

boo C3001... dwo ‘ ’3 ’J()c ..Jz. .‘D

goo pocomsr‘rr 7')

{as oocmcc:
yon 0100100579)

3‘21223»
-:~ 73;: car.) :7 :3
L’..;.t(_;- #930,;
”0"”09007030'0

' ‘ V
r

; ..JCIL‘) _‘~.)(‘ C‘.‘
L .1 poopq.'r."jc4
fittimnmmuw)n

 

 

!
|
3
I

-—- n-

'i
1:

  

 

i

- 1 .
’J 1 i ' L L,
vac: nooo ' .203 300435.29; Doc: DOD chfizoo oaoé- -o 1:, 37m;

co0.000.000.0090.0000080.oOOOOO'OOooodooioocapo-ow
22o one: one coat-.3330. 000 (23:13 cacao no 3 ooott-o.» 3:06.331?
Doo 0:30 .300 (.300 4.2.3:.) 0:30 :3 ‘30 00;: 330:: 000 on .3 uoaa‘
also cacao ooo soc: 0230 one coo cacao ‘ooo Doc: oooiéaom
age, coo moo concave 0:90 «.20: one «.300 Doc :oc‘c; oooq
44:0 000 ~D~D\D \D\D\D «(”0 003.0 0.00 @3030 'xDuhD 304330 \Dflxg \DchsL
umm‘mmm Nwmwwmm NNNWNNNNNNN NNN NNN NNN NNN Nmmm

'41 1 I l l I : l a

 

0.0.507

/

FTM
aurpu1;
Gt\

Eu:

S)'XL**?-t1

L
AJ'W‘UH)

”‘21NUUT3TAP
I)“

t—

   

PUT39UTPUT9TAP

..
I:

(‘le

 

VL'UHI
91‘)

‘18) S

,

c.
C)
if
0
9..
O
c:
A .
b
V "f
70
0

74/171)

I
n

 

new

EADGW
F( 305‘

1211‘

 

“V

R
I
Y
n

.DC ’.)L.. UC‘HJ) r- ...U‘

njfiKLf-i \lrflLC-
50C}

.... .

C

 

 

 

 

 

 

 

-LEMGTH*.8X¢"PRUPAGAIIOQ VELOCITY")

.2.ex,Fga.153

 

22.
. 5’5?

‘1~‘.)':)':‘
Ham)

 

 

GTION

0

1
LL
C.
Ll:
F.’

 

Appendix B. Standardization of G. C. Units

A.

105

 

 

 

 

 

0.17711. 0’5 Ia/a C0 MW wit.
-fk‘l
' ”‘EE 9%
J.
oé‘ao
I] ~“'
=
..j_
. STOP
RUN 3 4
anenz
RT RRER TYPE nR/HT nREaz
9.79 999999 9v 9.976 39.419
.g_r 123399 ve 9.937 19.581
10191 saga: 1213499
IUL rncron= 1.9999£+99
CO

 

 

 

PV = 7737'
(l4-Fa)(o./ x /0’3L;LAJ)(/5—zj: )7(0-032£%(29Y7()
M = 4-092 2!“?on co

)

Mm
m M 5.65%]:

4?
/2837O :: 3/87x/O 1“!ch

106

8. 047"! 0‘3 [00 PPM C01 Jig—«vMW-é;

 

J

J

L

l

llLlll

. 11
Va ‘Hrnr

 

7.84

UH I 7
..RER'I.

RT RRER TYPE flR/HT ARER'I.
z': 0699“ PB 8.878 98.357
W a? am 1643

 

0.4% <10; (WWW) 50:.

 

 

flZQ’ W7 M 7‘“!
pl/ 3 ”37'
C/ ”WNW! “0'34317/(16'4) = n (0-05’2;%)(271'fl)

77 = 7.637 x767 hate/$07,

)7, -7
WW’ 65C. 4%: 1.637 xIO//;4l5
zgrfx10‘4">..aeu col

H

107

C. 0'0, ”-2 ,3 90-f70 M— H7. ML
9..” n1 ‘5 79.9957. pure 6,”?

 

 

 

 

 

:ES
17
STOP
RUN O 3
GREAZ
RT GREQ TYPE flR/HT QRERZ
9.55 19817 PB 9.045 24.599
6:84 63744 88 0.071 75.481
TOTQL RRER= 88561
IUL FRCTOR= 1.0GGGE+80 1/
Z

 

 

6 «CZ-rm M
y M Aim/=flg7x/5’lmlflc Hz

 

 

 

 

 

 

 

:3;
_£:” ‘*"""‘ .84
[f 41.31
STOP
RUN 0 2
AREAZ
RT 6959 TYPE aR/HT AREAZ
0.84 66377 PB 0.677 8.641
1.31 638396 PB 6.114 91.359
TOTRL AREA: 698776
HUI. MOTOR: 1.6600900
c9;

 

~, .7
m Koo-7% = 6-lv‘lX/0—l hdfld CH+

Appendix C. Solutions to Equations (IV—l) to (IV-4)

108

M £0 5W (IV—/)Z.'o (IV—+2

g: X(§)YCC) (’9_1)

Mm 6W (J_/) Lara
EW‘ (IV—I).
.33: ” 2
X = -IY- = —)1 L 71>o)

(a) X”+ A2X=o

Xcg): AMZ’E-rBvaE
“p6,...” 5W (IV-24) 443.4 (IV-'26),
xci) :2 £44.92»;

_, 233*)
Ah“ T7? , M=oll;2)~s~

(b) y”- 73$: 0 7
my, = A’Manl + 8 m2»!
Wfi‘fip 5W («IV’ 2C) ,
A’=o

H400“, gC5,g-) = E AhM)H§W3h§
”’° (H72)

(IV-24) W MUA‘O‘v 74:02 W 7*"
W ”540147-420.
= E 14913170431371 Mani

ta! ”‘0

i ,4: M3135

”‘30

WA— fl:= flan» M371

wfi)
WHO)

109

-. a; . .
MM W5!!! ",5, Erin (’y-3) .

 

5
dz}; ‘0 ~ .
J z " 12,3 = $3 312-0 AMA-93»? (”’13)
3 3 .r
(«I g, = 5" + 5,.
5
£1. = «Mum + czw4>¢§

= CI 'y', C!) + C;f:(§7
£3? = “{"bjg +c+J}g<-}) a“! + fz(§)J§{‘ch’d’ Ji‘

 

 

 

 

 

   

     

 

 

 

 

W b WC-b
Mau—
Wd') :: ‘S'Mb {gab = _ #
HQ) igd) 2
”ma.
2;, = gay; 5‘. '2: 12......1mw1
1 ”=0
- Z'jfti ‘34,:2: AIM»? M+i A:
: haw
3' 42.449114 §Mh§~ )néfihgnghi]
(‘31:) ”'° 4',” + m,"
- z: ,, fiwqbzgmzn‘i- aways»:
(%)M¢;”:°A[: 4211-1-23."
= a A: 2»;
— ‘g ”z”; 41:: 33111
SS: CHM43-g +c7' W¢,§ .7 4353 ”1:32;:

W 5W LII/”+A} M CIV’4'A),

 

lo AM3n§
S'SC g) HA(W4>2§‘%M¢ MH€)— 3mm 4.11%»
(fl—4)

)7” WM 5W (fl—z) M (A_ +)AE
5W CIV-ZJ).

110

M m M
o = ..né; Anahmzamani +4;(.-§°A,ma2,.»2u§)

~++A<maf-M€M 4am- » gjfihjhfiffi»:

hdfimwm $7,“)..th . ¢

Ali—0,4 7W Etc £3 i=1,

"' é... Ananwah SIMRJMJME 1.3 + $. 50 “32.; a:
an o

-- 43% 79,..kaer ['aauthd A3
M4; M€§M2u§ ) 4?

 

(M4>,;443~3Ȥ '-

. I
’° finanmzh . .
+ €3¢+ {E E1" 2hz#( M23§M9n§ (£30

...thS

D

 

M muffi. ‘
J away ' .
hAMA”M2"[%"y’J¥f}'L + Air-M21”; .

 

 

 

 

 

"‘ «AMMAM[£_M~22§:]‘
z 4)». O I
" $4. [ fiw4’ng2h; - Amara-.433 m2%§{
+12 + 2)::- ‘0
~ I
- J’zwvhégMth— ahmfimzi’
'I‘Mda 4’2 “-45: + 2”?- p
Q ‘ ’
+ é AmfimJ—wfifim E - My)»; .
¢3 4’ $21. + 2,5. '2. _+))~T— 0 ... 0

flue $4»- ;9'». ,
A». = 2AJ>+ - 24>, c 451/5»)! +2)
M 2.4 434:? 4} 2.3) — 43mm” MJ/MH»)

MM ”’4'?”th 5W Cfl'Z),M-o£
)w’zo £=)—-E.

111

)7)“ , M M
'3. [ZAfi - .245, c Qua"- +I flu)”; w)”;-

”'° Mm ( +3 44' ' *3.“ 2):) " éW)n(%:/3m+w

S’: 1..
09-5)
L

”5‘ = '- L "elm A7

 

 

 

‘-"’— J... "" Dem C6 1%.!
L‘ a L. 3 ‘§=o L A; 1
ac
”DJ-Md N°/1=° "Dcm ax 170:0
= Dew-Co £2 2AA, - 24“ 45572.3“) _
L‘ h=° ¢3++ - ¢‘1_)”2_ 43(607‘ 2h)(fit/2H+®
New -—v 7
4 Na): :1 .. co <v> L

_ DCM do 2A4>4 ;24>.( éxz/Auz-Pl)

 

 

4 4’7 :-
L- m-o 4m * 42} “3:” 4: (mmM-z‘mw
_. 4),: a 094:)
4 x )5

 

~42in 4WARE-'4': 5"” +1 2:

leo‘b I! 5h,"
1

$44417, g z—zAlzcos/C'

W) = ”"w 0:704:71} “(Zn/Dw'w") L “(57095106

 

LIST OF REFERENCES

10.

11.

112

LIST OF REFERENCES

Cvetanoric, R.J., and Amenomiya, Y., "Application of a
Temperature-Programmed Desorption Technique to
Catalyst Studies," Advan. Catal. 11, 203 (1967)

Cvetanovic, R.J., and Amenomiya, Y., "A Temperature
Programmed Desorption Technique for Investigation
of Practical Catalysts," Catal. Rev. 6, 21 (1972)

Gorte, R.J., "Design Parameters for Temperature
Programmed Desorption from Porous Catalysts,"
J. Catal. 75, 164 (1982)

Hertz, R.K., Kiela, J.B., and Martin, S.P., "Adsorption
Effects during Temperature-Programmed Desorption of
Carbon Monoxide from Supported Platinum," J. Catal.
13, 66 (1982)

Foger, K., and Anderson, J.R., "Temperature Programmed
Desorption of Carbon Monoxide Adsorbed on Supported
Platinum Catalysts," Appl. Surf. Sci. 2, 335 (1979)

Taylor, J.L., and Weinburg, W.H., "A Method for Assessing
the Coverage Dependence of Kinetic Parameters: Appli—
cation to Carbon Monoxide Desorption from Iridium
(110)," Surface Sci. 28, 259 (1978)

Barton, 8.5., and Harrison, B.H., "Surface Studies on
Graphite: Desorptionof Surface Oxide," J. Chem. Soc.
Faraday Trans. I, 69, 1039 (1973)

Harker, H., Horsley, J.B., and Robson, D., "Active Centres
Produced in Graphite by Powdering," Carbon, 9, 1
(1971)

Mrozowski, S., and Andrew, J.F., "Electron Spin Resonance
of Broken Carbon Bonds," Proc. 4th Carbon Conf. 207
(Pergamon, Oxford, 1960)

Gorte, R., and Schmidt, L.D., "Temperature Programmed
Desorption with Reaction," Appl. Surface Sci. 3,
381 (1979)

Long, F.J., and Sykes, W.J., "Effect of Specific Catalysts
on the Reactions of the Steam—C System," J. Chem.
’Phys. 11, 361 (1950)

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

Long, F.J., and Sykes, W.J., "Effect of Specific Catalysts
on the Reactions of the Steam-C System," Prc. Royal
Soc. (London) A215, 100 (1952)

Walker, Jr., P.L., Shelef, M., and Anderson, R.A.,
"Catalysis of Carbon Gasification," in Chemistry and
Physics of Carbon, Vol. 4 (P.L. Walker, Jr. ed.),
Marcel Dekker, NEW York, 1968

 

 

Holstein, W.L., and Boudart, M., "Hydrogenolysis of Carbon
and its Catalysis by Platinum," J. Catal. 12, 51 (1981)

Holstein, W.L., and Boudart, M., "Transition Metal Oxide
Catalysed Gasification by Oxygen, Water, and Carbon
Dioxide," Fuel, 62, 162 (1983)

Baker, R.T.K., France, J.A., Rouse, L., and Waite, R.J.,
"Catalytic Oxidation of Graphite by Platinum and
Palladium," J. Catal. 41, 22 (1976)

Baker, R.T.K., Sherwood, R.D., "Catalytic Gasification of
Graphite by Nickel in Various Gaseous Environments,"
J. Catal. 10, 198 (1981)

Baker, R.T.K., Harris, P.S., Kemper, D.J., and Waite, R.J.,
"Controlled Atmosphere Electron Microscopy Studies of
Graphite Gasification - 3. The Catalytic Influence
of Molybdenum and Molybdenum Trioxide," Carbon 12,

179 (1974) '-

Baker, R.T.K., and Harris, P.S., "Controlled Atmosphere
Electron Microscopy Studies of Graphite Gasification
- 1. Catalytic Influence of Zinc," Carbon 14, 25
(1973) ——

Baker, R.T.K., and Chludzinski, Jr., J.J., "Catalytic
Gasification of Graphite by Chromium and Copper in
Oxygen, Steam, and Hydrogen," Carbon 19, 75 (1981)

Keep, C.W., Terry, 8., and Wells, M., "Studies of the
Nickel-Catalyzed Hydrogenation of Graphite," J. Catal.
66, 451 (1980)

Blakely, J.P., and Overholster, L.G., "Oxidation of ATJ
graphite by Low Concentrations of Water Vapor and
Carbon Dioxide in Helium," Carbon 3, 269 (1965)

Bond, G.C., Catalysis by Metals (Academic Press, New York,
1962)

 

24.

25.

26.

27.

28.

29.

30.

31.

32.

33,

34.

35.

36.

114

Feates, P.S., and Keep, C.W., "Surface Complexes formed
by Thermal and Radiolytic Adsorption of Gases on
Graphite," Trans. Faraday Soc., 66, 3156 (1970)

Thomas, J.M., Chemistry and Physics of Carbon, Vol. 1,
122 (P.L. Walker, Jr., ed.), Marcel Dekker, New York,
(1965)

 

Heintz, E.A., and Parker, W.E., "Catalytic Effect of Major
Impurities on Graphite Oxidation," Carbon 4, 473 (1966)

Winicur, D.H., "CO Desorption and Adsorption on Pt (111),"
Surface Sci. 109, 263 (1981)

Otto, K., and Shelef, M., "Catalytic Steam Gasification
of Graphite-Effects of Intercalated and Externally
Added Rum Rh, Pd and Pt," Carbon 15, 317 (1977)

Lalancette, J.M., U.S. Patent No. 3, 347, 963, dated
November 12, 1974

Alfa Division, Ventron Corporation Sales Catalog, p. 451
(1977-1978)

Otto, K., and Shelef, M., 6th Int. Congr. Catalysis, London,
1976, Paper B47.

McKee, D.W., "The Catalyzed Gasification Reactions of Carbon,‘
in Chemistry and Physics of Carbon (P.L. Waler, Jr.,
and P.A. Thrower, eds.) 16, 1 (1981), Marcel Dekker,
Inc., New York. -_

 

Feates, P.S., Harris, P.S., and Reuben, G.B., "Compensation
Effect in the Kinetics of the Catalyzed Oxidation of
Carbon," J. Chem. Soc. Faraday Trans. I, 70, 2011
(1974) "

Hennig, G.R., "Catalytic Oxidationof Graphite," J. Inorg.
Nucl. Chem. 34, 1129 (1962)

Presland, A.E.B., and Headley, J.A., "Electron-Microscope
Study of the Thermal Oxidationof Natrual Graphite,"
J. Nucl. Mater. 10, 99 (1963)

L'Homme, G.A., Boudart, M., and D'Or, L., "Kinetic Study of
Slow Oxidation of Channel Black by Gaseous Oxygen,"
Acad. Roy. Belg. Bull. Cl. Sci. 5 ser 52, 1249 (1966)

37.

38.

39.

40.

41.

42.

43.

44.

McKee, D.W., "Metal Oxides as Catalysts for the Oxidation
of Graphite," Carbon 8, 623 (1970)

Smith, D.J., Fisher, R.M., and Freeman, L.A., "High-
Resolution Electron Microscopy of Metal-Intercalated
"Graphimets"," J. Catal. 22, 51 (1981)

Magne, P., and Duval, X., "Existence of Four Types of
Reactive Sites in Graphite Oxidation," Bull. Soc.
Chim. Fr., A5, 1593 (1976)

Marsh, H., and Adair, R.R., "Catalytic Gasification of
Doped Carbon-A Kinetic Study," Carbon 13, 327 (1975)

Buch, T., Guala, J.A., and Caverio, A., "Oxyreactivity
of Doped Sucrose Carbon," Carbon 16, 377 (1978)

Polak, A., and Ehrlich, 6., "Surface Self-Diffusion,"
J. Vac. Sci. Technol. 14, 407 (1977)

L'nyanoi, V.N., "Determination of the Solid Solution-
Graphite Interfacial Energy in Nickel-Carbon, Cobalt-
Carbon, and Iron-Carbon Alloys," Fiz. Khim. Obrab.
Mater. 3, 104 (1977)

Ershov, G.S., and Kasatkin, A.A., "Effect of the Alloying
Elements on the Viscosity of Liquid Iron and Steels,"
Stal', 8, 712 (1977)